{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1643869394576601,
  "eval_steps": 375,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00010959129297177339,
      "grad_norm": 0.06448639929294586,
      "learning_rate": 2e-05,
      "loss": 10.3774,
      "step": 1
    },
    {
      "epoch": 0.00021918258594354678,
      "grad_norm": 0.06211783364415169,
      "learning_rate": 4e-05,
      "loss": 10.3763,
      "step": 2
    },
    {
      "epoch": 0.00032877387891532017,
      "grad_norm": 0.06450194865465164,
      "learning_rate": 6e-05,
      "loss": 10.3743,
      "step": 3
    },
    {
      "epoch": 0.00043836517188709356,
      "grad_norm": 0.06373259425163269,
      "learning_rate": 8e-05,
      "loss": 10.3764,
      "step": 4
    },
    {
      "epoch": 0.0005479564648588669,
      "grad_norm": 0.06279966980218887,
      "learning_rate": 0.0001,
      "loss": 10.3788,
      "step": 5
    },
    {
      "epoch": 0.0006575477578306403,
      "grad_norm": 0.06109637767076492,
      "learning_rate": 0.00012,
      "loss": 10.3796,
      "step": 6
    },
    {
      "epoch": 0.0007671390508024137,
      "grad_norm": 0.06335120648145676,
      "learning_rate": 0.00014,
      "loss": 10.3786,
      "step": 7
    },
    {
      "epoch": 0.0008767303437741871,
      "grad_norm": 0.05948799103498459,
      "learning_rate": 0.00016,
      "loss": 10.3756,
      "step": 8
    },
    {
      "epoch": 0.0009863216367459605,
      "grad_norm": 0.06671092659235,
      "learning_rate": 0.00018,
      "loss": 10.375,
      "step": 9
    },
    {
      "epoch": 0.0010959129297177339,
      "grad_norm": 0.07076307386159897,
      "learning_rate": 0.0002,
      "loss": 10.3781,
      "step": 10
    },
    {
      "epoch": 0.0012055042226895073,
      "grad_norm": 0.0715995579957962,
      "learning_rate": 0.00019999977772170748,
      "loss": 10.3808,
      "step": 11
    },
    {
      "epoch": 0.0013150955156612807,
      "grad_norm": 0.0610898919403553,
      "learning_rate": 0.00019999911088781805,
      "loss": 10.3811,
      "step": 12
    },
    {
      "epoch": 0.001424686808633054,
      "grad_norm": 0.06454722583293915,
      "learning_rate": 0.0001999979995012962,
      "loss": 10.3753,
      "step": 13
    },
    {
      "epoch": 0.0015342781016048274,
      "grad_norm": 0.06984736025333405,
      "learning_rate": 0.00019999644356708261,
      "loss": 10.3765,
      "step": 14
    },
    {
      "epoch": 0.0016438693945766008,
      "grad_norm": 0.06784087419509888,
      "learning_rate": 0.00019999444309209432,
      "loss": 10.3733,
      "step": 15
    },
    {
      "epoch": 0.0017534606875483742,
      "grad_norm": 0.07476300001144409,
      "learning_rate": 0.0001999919980852246,
      "loss": 10.3752,
      "step": 16
    },
    {
      "epoch": 0.0018630519805201476,
      "grad_norm": 0.07227816432714462,
      "learning_rate": 0.00019998910855734288,
      "loss": 10.3718,
      "step": 17
    },
    {
      "epoch": 0.001972643273491921,
      "grad_norm": 0.07505158334970474,
      "learning_rate": 0.0001999857745212947,
      "loss": 10.3759,
      "step": 18
    },
    {
      "epoch": 0.0020822345664636946,
      "grad_norm": 0.08245861530303955,
      "learning_rate": 0.00019998199599190178,
      "loss": 10.3736,
      "step": 19
    },
    {
      "epoch": 0.0021918258594354678,
      "grad_norm": 0.08533550053834915,
      "learning_rate": 0.0001999777729859618,
      "loss": 10.3694,
      "step": 20
    },
    {
      "epoch": 0.0023014171524072414,
      "grad_norm": 0.08868667483329773,
      "learning_rate": 0.00019997310552224846,
      "loss": 10.3733,
      "step": 21
    },
    {
      "epoch": 0.0024110084453790146,
      "grad_norm": 0.084206722676754,
      "learning_rate": 0.00019996799362151122,
      "loss": 10.3716,
      "step": 22
    },
    {
      "epoch": 0.002520599738350788,
      "grad_norm": 0.08865362405776978,
      "learning_rate": 0.00019996243730647538,
      "loss": 10.3716,
      "step": 23
    },
    {
      "epoch": 0.0026301910313225613,
      "grad_norm": 0.10592816025018692,
      "learning_rate": 0.00019995643660184191,
      "loss": 10.3694,
      "step": 24
    },
    {
      "epoch": 0.002739782324294335,
      "grad_norm": 0.09898102283477783,
      "learning_rate": 0.00019994999153428737,
      "loss": 10.3701,
      "step": 25
    },
    {
      "epoch": 0.002849373617266108,
      "grad_norm": 0.07873634248971939,
      "learning_rate": 0.00019994310213246368,
      "loss": 10.3683,
      "step": 26
    },
    {
      "epoch": 0.0029589649102378817,
      "grad_norm": 0.10535549372434616,
      "learning_rate": 0.00019993576842699816,
      "loss": 10.3705,
      "step": 27
    },
    {
      "epoch": 0.003068556203209655,
      "grad_norm": 0.1002214178442955,
      "learning_rate": 0.0001999279904504933,
      "loss": 10.3673,
      "step": 28
    },
    {
      "epoch": 0.0031781474961814285,
      "grad_norm": 0.10811527073383331,
      "learning_rate": 0.00019991976823752653,
      "loss": 10.3685,
      "step": 29
    },
    {
      "epoch": 0.0032877387891532017,
      "grad_norm": 0.10568730533123016,
      "learning_rate": 0.00019991110182465032,
      "loss": 10.3702,
      "step": 30
    },
    {
      "epoch": 0.0033973300821249753,
      "grad_norm": 0.12007065117359161,
      "learning_rate": 0.00019990199125039174,
      "loss": 10.3732,
      "step": 31
    },
    {
      "epoch": 0.0035069213750967484,
      "grad_norm": 0.13536852598190308,
      "learning_rate": 0.00019989243655525247,
      "loss": 10.3601,
      "step": 32
    },
    {
      "epoch": 0.003616512668068522,
      "grad_norm": 0.11974607408046722,
      "learning_rate": 0.00019988243778170853,
      "loss": 10.3663,
      "step": 33
    },
    {
      "epoch": 0.0037261039610402952,
      "grad_norm": 0.13050992786884308,
      "learning_rate": 0.0001998719949742101,
      "loss": 10.3593,
      "step": 34
    },
    {
      "epoch": 0.003835695254012069,
      "grad_norm": 0.13830652832984924,
      "learning_rate": 0.0001998611081791814,
      "loss": 10.3625,
      "step": 35
    },
    {
      "epoch": 0.003945286546983842,
      "grad_norm": 0.14223182201385498,
      "learning_rate": 0.00019984977744502038,
      "loss": 10.3617,
      "step": 36
    },
    {
      "epoch": 0.004054877839955616,
      "grad_norm": 0.15017090737819672,
      "learning_rate": 0.00019983800282209857,
      "loss": 10.3612,
      "step": 37
    },
    {
      "epoch": 0.004164469132927389,
      "grad_norm": 0.13818952441215515,
      "learning_rate": 0.00019982578436276082,
      "loss": 10.3594,
      "step": 38
    },
    {
      "epoch": 0.004274060425899162,
      "grad_norm": 0.14971062541007996,
      "learning_rate": 0.00019981312212132512,
      "loss": 10.3558,
      "step": 39
    },
    {
      "epoch": 0.0043836517188709356,
      "grad_norm": 0.16297782957553864,
      "learning_rate": 0.00019980001615408228,
      "loss": 10.355,
      "step": 40
    },
    {
      "epoch": 0.004493243011842709,
      "grad_norm": 0.16278710961341858,
      "learning_rate": 0.00019978646651929572,
      "loss": 10.3593,
      "step": 41
    },
    {
      "epoch": 0.004602834304814483,
      "grad_norm": 0.17110323905944824,
      "learning_rate": 0.00019977247327720128,
      "loss": 10.3531,
      "step": 42
    },
    {
      "epoch": 0.0047124255977862555,
      "grad_norm": 0.179261714220047,
      "learning_rate": 0.0001997580364900068,
      "loss": 10.3566,
      "step": 43
    },
    {
      "epoch": 0.004822016890758029,
      "grad_norm": 0.18145814538002014,
      "learning_rate": 0.000199743156221892,
      "loss": 10.3535,
      "step": 44
    },
    {
      "epoch": 0.004931608183729803,
      "grad_norm": 0.18181148171424866,
      "learning_rate": 0.00019972783253900808,
      "loss": 10.3538,
      "step": 45
    },
    {
      "epoch": 0.005041199476701576,
      "grad_norm": 0.205839142203331,
      "learning_rate": 0.00019971206550947748,
      "loss": 10.3456,
      "step": 46
    },
    {
      "epoch": 0.005150790769673349,
      "grad_norm": 0.1862969845533371,
      "learning_rate": 0.00019969585520339354,
      "loss": 10.3491,
      "step": 47
    },
    {
      "epoch": 0.005260382062645123,
      "grad_norm": 0.18707241117954254,
      "learning_rate": 0.0001996792016928203,
      "loss": 10.3548,
      "step": 48
    },
    {
      "epoch": 0.005369973355616896,
      "grad_norm": 0.18767386674880981,
      "learning_rate": 0.00019966210505179197,
      "loss": 10.3461,
      "step": 49
    },
    {
      "epoch": 0.00547956464858867,
      "grad_norm": 0.18882805109024048,
      "learning_rate": 0.00019964456535631286,
      "loss": 10.3445,
      "step": 50
    },
    {
      "epoch": 0.005589155941560443,
      "grad_norm": 0.1994628757238388,
      "learning_rate": 0.0001996265826843568,
      "loss": 10.3428,
      "step": 51
    },
    {
      "epoch": 0.005698747234532216,
      "grad_norm": 0.1827801913022995,
      "learning_rate": 0.00019960815711586696,
      "loss": 10.3435,
      "step": 52
    },
    {
      "epoch": 0.00580833852750399,
      "grad_norm": 0.19601450860500336,
      "learning_rate": 0.00019958928873275539,
      "loss": 10.3423,
      "step": 53
    },
    {
      "epoch": 0.0059179298204757634,
      "grad_norm": 0.19590751826763153,
      "learning_rate": 0.00019956997761890277,
      "loss": 10.3394,
      "step": 54
    },
    {
      "epoch": 0.006027521113447536,
      "grad_norm": 0.20690293610095978,
      "learning_rate": 0.00019955022386015792,
      "loss": 10.3357,
      "step": 55
    },
    {
      "epoch": 0.00613711240641931,
      "grad_norm": 0.19882230460643768,
      "learning_rate": 0.00019953002754433743,
      "loss": 10.3307,
      "step": 56
    },
    {
      "epoch": 0.006246703699391083,
      "grad_norm": 0.19620352983474731,
      "learning_rate": 0.00019950938876122542,
      "loss": 10.3279,
      "step": 57
    },
    {
      "epoch": 0.006356294992362857,
      "grad_norm": 0.1937166005373001,
      "learning_rate": 0.00019948830760257291,
      "loss": 10.3291,
      "step": 58
    },
    {
      "epoch": 0.00646588628533463,
      "grad_norm": 0.18030841648578644,
      "learning_rate": 0.0001994667841620976,
      "loss": 10.3264,
      "step": 59
    },
    {
      "epoch": 0.006575477578306403,
      "grad_norm": 0.18191669881343842,
      "learning_rate": 0.00019944481853548335,
      "loss": 10.3271,
      "step": 60
    },
    {
      "epoch": 0.006685068871278177,
      "grad_norm": 0.1941377967596054,
      "learning_rate": 0.00019942241082037982,
      "loss": 10.3193,
      "step": 61
    },
    {
      "epoch": 0.0067946601642499506,
      "grad_norm": 0.1746060997247696,
      "learning_rate": 0.00019939956111640197,
      "loss": 10.3237,
      "step": 62
    },
    {
      "epoch": 0.006904251457221724,
      "grad_norm": 0.1629510521888733,
      "learning_rate": 0.00019937626952512964,
      "loss": 10.3233,
      "step": 63
    },
    {
      "epoch": 0.007013842750193497,
      "grad_norm": 0.16537755727767944,
      "learning_rate": 0.0001993525361501072,
      "loss": 10.3237,
      "step": 64
    },
    {
      "epoch": 0.0071234340431652705,
      "grad_norm": 0.1657216101884842,
      "learning_rate": 0.00019932836109684286,
      "loss": 10.3122,
      "step": 65
    },
    {
      "epoch": 0.007233025336137044,
      "grad_norm": 0.1552131325006485,
      "learning_rate": 0.00019930374447280845,
      "loss": 10.3165,
      "step": 66
    },
    {
      "epoch": 0.007342616629108818,
      "grad_norm": 0.14066174626350403,
      "learning_rate": 0.00019927868638743875,
      "loss": 10.3155,
      "step": 67
    },
    {
      "epoch": 0.0074522079220805905,
      "grad_norm": 0.15565122663974762,
      "learning_rate": 0.0001992531869521312,
      "loss": 10.3108,
      "step": 68
    },
    {
      "epoch": 0.007561799215052364,
      "grad_norm": 0.1355278044939041,
      "learning_rate": 0.00019922724628024515,
      "loss": 10.3187,
      "step": 69
    },
    {
      "epoch": 0.007671390508024138,
      "grad_norm": 0.12752875685691833,
      "learning_rate": 0.0001992008644871016,
      "loss": 10.318,
      "step": 70
    },
    {
      "epoch": 0.007780981800995911,
      "grad_norm": 0.13228414952754974,
      "learning_rate": 0.00019917404168998256,
      "loss": 10.3107,
      "step": 71
    },
    {
      "epoch": 0.007890573093967684,
      "grad_norm": 0.1377786099910736,
      "learning_rate": 0.0001991467780081305,
      "loss": 10.3083,
      "step": 72
    },
    {
      "epoch": 0.008000164386939458,
      "grad_norm": 0.1352386474609375,
      "learning_rate": 0.00019911907356274795,
      "loss": 10.3101,
      "step": 73
    },
    {
      "epoch": 0.008109755679911231,
      "grad_norm": 0.11667723953723907,
      "learning_rate": 0.00019909092847699683,
      "loss": 10.3108,
      "step": 74
    },
    {
      "epoch": 0.008219346972883004,
      "grad_norm": 0.12652815878391266,
      "learning_rate": 0.00019906234287599798,
      "loss": 10.3119,
      "step": 75
    },
    {
      "epoch": 0.008328938265854778,
      "grad_norm": 0.13222317397594452,
      "learning_rate": 0.00019903331688683057,
      "loss": 10.3043,
      "step": 76
    },
    {
      "epoch": 0.008438529558826551,
      "grad_norm": 0.12171619385480881,
      "learning_rate": 0.00019900385063853154,
      "loss": 10.3031,
      "step": 77
    },
    {
      "epoch": 0.008548120851798324,
      "grad_norm": 0.11676570773124695,
      "learning_rate": 0.00019897394426209505,
      "loss": 10.3085,
      "step": 78
    },
    {
      "epoch": 0.008657712144770098,
      "grad_norm": 0.1004306823015213,
      "learning_rate": 0.00019894359789047187,
      "loss": 10.3028,
      "step": 79
    },
    {
      "epoch": 0.008767303437741871,
      "grad_norm": 0.12803275883197784,
      "learning_rate": 0.00019891281165856873,
      "loss": 10.302,
      "step": 80
    },
    {
      "epoch": 0.008876894730713646,
      "grad_norm": 0.09693340957164764,
      "learning_rate": 0.00019888158570324795,
      "loss": 10.3048,
      "step": 81
    },
    {
      "epoch": 0.008986486023685418,
      "grad_norm": 0.10417886078357697,
      "learning_rate": 0.0001988499201633265,
      "loss": 10.3104,
      "step": 82
    },
    {
      "epoch": 0.009096077316657191,
      "grad_norm": 0.11355624347925186,
      "learning_rate": 0.00019881781517957562,
      "loss": 10.3045,
      "step": 83
    },
    {
      "epoch": 0.009205668609628966,
      "grad_norm": 0.1104326918721199,
      "learning_rate": 0.0001987852708947202,
      "loss": 10.2972,
      "step": 84
    },
    {
      "epoch": 0.009315259902600738,
      "grad_norm": 0.10576529055833817,
      "learning_rate": 0.00019875228745343794,
      "loss": 10.2957,
      "step": 85
    },
    {
      "epoch": 0.009424851195572511,
      "grad_norm": 0.09020967036485672,
      "learning_rate": 0.0001987188650023589,
      "loss": 10.2998,
      "step": 86
    },
    {
      "epoch": 0.009534442488544285,
      "grad_norm": 0.10750852525234222,
      "learning_rate": 0.0001986850036900648,
      "loss": 10.3044,
      "step": 87
    },
    {
      "epoch": 0.009644033781516058,
      "grad_norm": 0.09546364843845367,
      "learning_rate": 0.00019865070366708836,
      "loss": 10.3017,
      "step": 88
    },
    {
      "epoch": 0.009753625074487833,
      "grad_norm": 0.09333126246929169,
      "learning_rate": 0.00019861596508591255,
      "loss": 10.2976,
      "step": 89
    },
    {
      "epoch": 0.009863216367459605,
      "grad_norm": 0.09172912687063217,
      "learning_rate": 0.00019858078810097002,
      "loss": 10.3025,
      "step": 90
    },
    {
      "epoch": 0.009972807660431378,
      "grad_norm": 0.08087832480669022,
      "learning_rate": 0.00019854517286864245,
      "loss": 10.297,
      "step": 91
    },
    {
      "epoch": 0.010082398953403153,
      "grad_norm": 0.08962374180555344,
      "learning_rate": 0.0001985091195472596,
      "loss": 10.2993,
      "step": 92
    },
    {
      "epoch": 0.010191990246374925,
      "grad_norm": 0.0777488499879837,
      "learning_rate": 0.0001984726282970989,
      "loss": 10.3073,
      "step": 93
    },
    {
      "epoch": 0.010301581539346698,
      "grad_norm": 0.0805872231721878,
      "learning_rate": 0.0001984356992803847,
      "loss": 10.2961,
      "step": 94
    },
    {
      "epoch": 0.010411172832318473,
      "grad_norm": 0.08461452275514603,
      "learning_rate": 0.00019839833266128724,
      "loss": 10.3048,
      "step": 95
    },
    {
      "epoch": 0.010520764125290245,
      "grad_norm": 0.07459169626235962,
      "learning_rate": 0.00019836052860592237,
      "loss": 10.2997,
      "step": 96
    },
    {
      "epoch": 0.01063035541826202,
      "grad_norm": 0.07693763077259064,
      "learning_rate": 0.0001983222872823505,
      "loss": 10.3018,
      "step": 97
    },
    {
      "epoch": 0.010739946711233793,
      "grad_norm": 0.08199900388717651,
      "learning_rate": 0.00019828360886057594,
      "loss": 10.3,
      "step": 98
    },
    {
      "epoch": 0.010849538004205565,
      "grad_norm": 0.080073781311512,
      "learning_rate": 0.00019824449351254616,
      "loss": 10.3027,
      "step": 99
    },
    {
      "epoch": 0.01095912929717734,
      "grad_norm": 0.09909723699092865,
      "learning_rate": 0.00019820494141215104,
      "loss": 10.3121,
      "step": 100
    },
    {
      "epoch": 0.011068720590149113,
      "grad_norm": 0.06665422767400742,
      "learning_rate": 0.000198164952735222,
      "loss": 10.3031,
      "step": 101
    },
    {
      "epoch": 0.011178311883120885,
      "grad_norm": 0.06412188708782196,
      "learning_rate": 0.00019812452765953135,
      "loss": 10.3062,
      "step": 102
    },
    {
      "epoch": 0.01128790317609266,
      "grad_norm": 0.07259787619113922,
      "learning_rate": 0.00019808366636479147,
      "loss": 10.3009,
      "step": 103
    },
    {
      "epoch": 0.011397494469064432,
      "grad_norm": 0.0607011541724205,
      "learning_rate": 0.00019804236903265388,
      "loss": 10.301,
      "step": 104
    },
    {
      "epoch": 0.011507085762036207,
      "grad_norm": 0.05484991520643234,
      "learning_rate": 0.00019800063584670863,
      "loss": 10.3045,
      "step": 105
    },
    {
      "epoch": 0.01161667705500798,
      "grad_norm": 0.05612761899828911,
      "learning_rate": 0.00019795846699248332,
      "loss": 10.2987,
      "step": 106
    },
    {
      "epoch": 0.011726268347979752,
      "grad_norm": 0.06183931604027748,
      "learning_rate": 0.00019791586265744237,
      "loss": 10.3043,
      "step": 107
    },
    {
      "epoch": 0.011835859640951527,
      "grad_norm": 0.06866899132728577,
      "learning_rate": 0.00019787282303098617,
      "loss": 10.2952,
      "step": 108
    },
    {
      "epoch": 0.0119454509339233,
      "grad_norm": 0.06475495547056198,
      "learning_rate": 0.0001978293483044502,
      "loss": 10.2953,
      "step": 109
    },
    {
      "epoch": 0.012055042226895072,
      "grad_norm": 0.05250517278909683,
      "learning_rate": 0.00019778543867110426,
      "loss": 10.3073,
      "step": 110
    },
    {
      "epoch": 0.012164633519866847,
      "grad_norm": 0.05306226760149002,
      "learning_rate": 0.00019774109432615147,
      "loss": 10.3045,
      "step": 111
    },
    {
      "epoch": 0.01227422481283862,
      "grad_norm": 0.07671590149402618,
      "learning_rate": 0.00019769631546672756,
      "loss": 10.2973,
      "step": 112
    },
    {
      "epoch": 0.012383816105810394,
      "grad_norm": 0.06117364019155502,
      "learning_rate": 0.00019765110229189988,
      "loss": 10.3081,
      "step": 113
    },
    {
      "epoch": 0.012493407398782167,
      "grad_norm": 0.0759350135922432,
      "learning_rate": 0.00019760545500266657,
      "loss": 10.3005,
      "step": 114
    },
    {
      "epoch": 0.01260299869175394,
      "grad_norm": 0.06545155495405197,
      "learning_rate": 0.00019755937380195568,
      "loss": 10.3037,
      "step": 115
    },
    {
      "epoch": 0.012712589984725714,
      "grad_norm": 0.0696941614151001,
      "learning_rate": 0.00019751285889462423,
      "loss": 10.3102,
      "step": 116
    },
    {
      "epoch": 0.012822181277697487,
      "grad_norm": 0.057997167110443115,
      "learning_rate": 0.0001974659104874573,
      "loss": 10.3025,
      "step": 117
    },
    {
      "epoch": 0.01293177257066926,
      "grad_norm": 0.0548679418861866,
      "learning_rate": 0.0001974185287891671,
      "loss": 10.2944,
      "step": 118
    },
    {
      "epoch": 0.013041363863641034,
      "grad_norm": 0.054358530789613724,
      "learning_rate": 0.0001973707140103921,
      "loss": 10.2901,
      "step": 119
    },
    {
      "epoch": 0.013150955156612807,
      "grad_norm": 0.05729759484529495,
      "learning_rate": 0.00019732246636369605,
      "loss": 10.2981,
      "step": 120
    },
    {
      "epoch": 0.013260546449584581,
      "grad_norm": 0.0534212663769722,
      "learning_rate": 0.00019727378606356703,
      "loss": 10.3029,
      "step": 121
    },
    {
      "epoch": 0.013370137742556354,
      "grad_norm": 0.05572551488876343,
      "learning_rate": 0.00019722467332641656,
      "loss": 10.2947,
      "step": 122
    },
    {
      "epoch": 0.013479729035528127,
      "grad_norm": 0.06268574297428131,
      "learning_rate": 0.00019717512837057855,
      "loss": 10.2986,
      "step": 123
    },
    {
      "epoch": 0.013589320328499901,
      "grad_norm": 0.07702918350696564,
      "learning_rate": 0.0001971251514163083,
      "loss": 10.2845,
      "step": 124
    },
    {
      "epoch": 0.013698911621471674,
      "grad_norm": 0.06270226836204529,
      "learning_rate": 0.0001970747426857817,
      "loss": 10.291,
      "step": 125
    },
    {
      "epoch": 0.013808502914443448,
      "grad_norm": 0.06534460932016373,
      "learning_rate": 0.00019702390240309404,
      "loss": 10.2952,
      "step": 126
    },
    {
      "epoch": 0.013918094207415221,
      "grad_norm": 0.058862052857875824,
      "learning_rate": 0.0001969726307942592,
      "loss": 10.2978,
      "step": 127
    },
    {
      "epoch": 0.014027685500386994,
      "grad_norm": 0.05975919961929321,
      "learning_rate": 0.00019692092808720846,
      "loss": 10.2994,
      "step": 128
    },
    {
      "epoch": 0.014137276793358768,
      "grad_norm": 0.06481282413005829,
      "learning_rate": 0.0001968687945117896,
      "loss": 10.299,
      "step": 129
    },
    {
      "epoch": 0.014246868086330541,
      "grad_norm": 0.06619228422641754,
      "learning_rate": 0.00019681623029976588,
      "loss": 10.2859,
      "step": 130
    },
    {
      "epoch": 0.014356459379302314,
      "grad_norm": 0.0694047212600708,
      "learning_rate": 0.00019676323568481498,
      "loss": 10.2914,
      "step": 131
    },
    {
      "epoch": 0.014466050672274088,
      "grad_norm": 0.05629447102546692,
      "learning_rate": 0.00019670981090252792,
      "loss": 10.3,
      "step": 132
    },
    {
      "epoch": 0.014575641965245861,
      "grad_norm": 0.060504477471113205,
      "learning_rate": 0.00019665595619040808,
      "loss": 10.2954,
      "step": 133
    },
    {
      "epoch": 0.014685233258217635,
      "grad_norm": 0.0583382174372673,
      "learning_rate": 0.0001966016717878702,
      "loss": 10.3026,
      "step": 134
    },
    {
      "epoch": 0.014794824551189408,
      "grad_norm": 0.07208079099655151,
      "learning_rate": 0.00019654695793623907,
      "loss": 10.2872,
      "step": 135
    },
    {
      "epoch": 0.014904415844161181,
      "grad_norm": 0.08228544890880585,
      "learning_rate": 0.0001964918148787488,
      "loss": 10.2838,
      "step": 136
    },
    {
      "epoch": 0.015014007137132955,
      "grad_norm": 0.06522556394338608,
      "learning_rate": 0.00019643624286054144,
      "loss": 10.2944,
      "step": 137
    },
    {
      "epoch": 0.015123598430104728,
      "grad_norm": 0.05958382785320282,
      "learning_rate": 0.00019638024212866606,
      "loss": 10.3008,
      "step": 138
    },
    {
      "epoch": 0.0152331897230765,
      "grad_norm": 0.07364761084318161,
      "learning_rate": 0.0001963238129320776,
      "loss": 10.2904,
      "step": 139
    },
    {
      "epoch": 0.015342781016048275,
      "grad_norm": 0.08039070665836334,
      "learning_rate": 0.00019626695552163578,
      "loss": 10.299,
      "step": 140
    },
    {
      "epoch": 0.015452372309020048,
      "grad_norm": 0.06334484368562698,
      "learning_rate": 0.00019620967015010395,
      "loss": 10.2973,
      "step": 141
    },
    {
      "epoch": 0.015561963601991823,
      "grad_norm": 0.0630512610077858,
      "learning_rate": 0.00019615195707214803,
      "loss": 10.2981,
      "step": 142
    },
    {
      "epoch": 0.015671554894963594,
      "grad_norm": 0.06689666211605072,
      "learning_rate": 0.0001960938165443353,
      "loss": 10.2951,
      "step": 143
    },
    {
      "epoch": 0.015781146187935368,
      "grad_norm": 0.08282075077295303,
      "learning_rate": 0.00019603524882513327,
      "loss": 10.293,
      "step": 144
    },
    {
      "epoch": 0.015890737480907142,
      "grad_norm": 0.06361265480518341,
      "learning_rate": 0.0001959762541749086,
      "loss": 10.2902,
      "step": 145
    },
    {
      "epoch": 0.016000328773878917,
      "grad_norm": 0.06523275375366211,
      "learning_rate": 0.00019591683285592593,
      "loss": 10.3022,
      "step": 146
    },
    {
      "epoch": 0.016109920066850688,
      "grad_norm": 0.06948568671941757,
      "learning_rate": 0.00019585698513234663,
      "loss": 10.2968,
      "step": 147
    },
    {
      "epoch": 0.016219511359822462,
      "grad_norm": 0.06411261111497879,
      "learning_rate": 0.0001957967112702277,
      "loss": 10.2941,
      "step": 148
    },
    {
      "epoch": 0.016329102652794237,
      "grad_norm": 0.0811004489660263,
      "learning_rate": 0.00019573601153752052,
      "loss": 10.2919,
      "step": 149
    },
    {
      "epoch": 0.016438693945766008,
      "grad_norm": 0.07777290791273117,
      "learning_rate": 0.00019567488620406983,
      "loss": 10.2996,
      "step": 150
    },
    {
      "epoch": 0.016548285238737782,
      "grad_norm": 0.05361684784293175,
      "learning_rate": 0.00019561333554161224,
      "loss": 10.2956,
      "step": 151
    },
    {
      "epoch": 0.016657876531709557,
      "grad_norm": 0.05512464419007301,
      "learning_rate": 0.0001955513598237753,
      "loss": 10.2919,
      "step": 152
    },
    {
      "epoch": 0.016767467824681328,
      "grad_norm": 0.060840439051389694,
      "learning_rate": 0.00019548895932607621,
      "loss": 10.2924,
      "step": 153
    },
    {
      "epoch": 0.016877059117653102,
      "grad_norm": 0.06367655098438263,
      "learning_rate": 0.00019542613432592038,
      "loss": 10.2978,
      "step": 154
    },
    {
      "epoch": 0.016986650410624877,
      "grad_norm": 0.04754406213760376,
      "learning_rate": 0.00019536288510260056,
      "loss": 10.2972,
      "step": 155
    },
    {
      "epoch": 0.017096241703596648,
      "grad_norm": 0.05843094363808632,
      "learning_rate": 0.00019529921193729534,
      "loss": 10.2977,
      "step": 156
    },
    {
      "epoch": 0.017205832996568422,
      "grad_norm": 0.0646003782749176,
      "learning_rate": 0.00019523511511306793,
      "loss": 10.2957,
      "step": 157
    },
    {
      "epoch": 0.017315424289540197,
      "grad_norm": 0.053422972559928894,
      "learning_rate": 0.000195170594914865,
      "loss": 10.2982,
      "step": 158
    },
    {
      "epoch": 0.017425015582511968,
      "grad_norm": 0.05492417886853218,
      "learning_rate": 0.00019510565162951537,
      "loss": 10.2955,
      "step": 159
    },
    {
      "epoch": 0.017534606875483742,
      "grad_norm": 0.06162506714463234,
      "learning_rate": 0.00019504028554572864,
      "loss": 10.3005,
      "step": 160
    },
    {
      "epoch": 0.017644198168455517,
      "grad_norm": 0.05936793237924576,
      "learning_rate": 0.00019497449695409408,
      "loss": 10.2932,
      "step": 161
    },
    {
      "epoch": 0.01775378946142729,
      "grad_norm": 0.05899824947118759,
      "learning_rate": 0.00019490828614707916,
      "loss": 10.2987,
      "step": 162
    },
    {
      "epoch": 0.017863380754399062,
      "grad_norm": 0.060677286237478256,
      "learning_rate": 0.00019484165341902845,
      "loss": 10.289,
      "step": 163
    },
    {
      "epoch": 0.017972972047370837,
      "grad_norm": 0.057306528091430664,
      "learning_rate": 0.00019477459906616206,
      "loss": 10.2974,
      "step": 164
    },
    {
      "epoch": 0.01808256334034261,
      "grad_norm": 0.06600688397884369,
      "learning_rate": 0.00019470712338657458,
      "loss": 10.2973,
      "step": 165
    },
    {
      "epoch": 0.018192154633314382,
      "grad_norm": 0.05563896521925926,
      "learning_rate": 0.0001946392266802336,
      "loss": 10.2994,
      "step": 166
    },
    {
      "epoch": 0.018301745926286157,
      "grad_norm": 0.05156634747982025,
      "learning_rate": 0.0001945709092489783,
      "loss": 10.2922,
      "step": 167
    },
    {
      "epoch": 0.01841133721925793,
      "grad_norm": 0.07582421600818634,
      "learning_rate": 0.00019450217139651844,
      "loss": 10.3034,
      "step": 168
    },
    {
      "epoch": 0.018520928512229702,
      "grad_norm": 0.05762793496251106,
      "learning_rate": 0.0001944330134284326,
      "loss": 10.2977,
      "step": 169
    },
    {
      "epoch": 0.018630519805201477,
      "grad_norm": 0.05056002363562584,
      "learning_rate": 0.00019436343565216711,
      "loss": 10.2953,
      "step": 170
    },
    {
      "epoch": 0.01874011109817325,
      "grad_norm": 0.05217313393950462,
      "learning_rate": 0.00019429343837703455,
      "loss": 10.2931,
      "step": 171
    },
    {
      "epoch": 0.018849702391145022,
      "grad_norm": 0.05761442333459854,
      "learning_rate": 0.0001942230219142124,
      "loss": 10.2991,
      "step": 172
    },
    {
      "epoch": 0.018959293684116797,
      "grad_norm": 0.060214802622795105,
      "learning_rate": 0.0001941521865767417,
      "loss": 10.2941,
      "step": 173
    },
    {
      "epoch": 0.01906888497708857,
      "grad_norm": 0.05381496623158455,
      "learning_rate": 0.0001940809326795256,
      "loss": 10.2919,
      "step": 174
    },
    {
      "epoch": 0.019178476270060342,
      "grad_norm": 0.05470126122236252,
      "learning_rate": 0.000194009260539328,
      "loss": 10.2926,
      "step": 175
    },
    {
      "epoch": 0.019288067563032116,
      "grad_norm": 0.053889598697423935,
      "learning_rate": 0.0001939371704747721,
      "loss": 10.2892,
      "step": 176
    },
    {
      "epoch": 0.01939765885600389,
      "grad_norm": 0.04970921576023102,
      "learning_rate": 0.00019386466280633906,
      "loss": 10.3014,
      "step": 177
    },
    {
      "epoch": 0.019507250148975665,
      "grad_norm": 0.04187307879328728,
      "learning_rate": 0.00019379173785636646,
      "loss": 10.302,
      "step": 178
    },
    {
      "epoch": 0.019616841441947436,
      "grad_norm": 0.061843741685152054,
      "learning_rate": 0.000193718395949047,
      "loss": 10.2835,
      "step": 179
    },
    {
      "epoch": 0.01972643273491921,
      "grad_norm": 0.05556268244981766,
      "learning_rate": 0.00019364463741042694,
      "loss": 10.288,
      "step": 180
    },
    {
      "epoch": 0.019836024027890985,
      "grad_norm": 0.05342453718185425,
      "learning_rate": 0.00019357046256840473,
      "loss": 10.2934,
      "step": 181
    },
    {
      "epoch": 0.019945615320862756,
      "grad_norm": 0.06199873611330986,
      "learning_rate": 0.00019349587175272948,
      "loss": 10.2968,
      "step": 182
    },
    {
      "epoch": 0.02005520661383453,
      "grad_norm": 0.04670555144548416,
      "learning_rate": 0.0001934208652949996,
      "loss": 10.2948,
      "step": 183
    },
    {
      "epoch": 0.020164797906806305,
      "grad_norm": 0.06348498910665512,
      "learning_rate": 0.00019334544352866127,
      "loss": 10.3018,
      "step": 184
    },
    {
      "epoch": 0.020274389199778076,
      "grad_norm": 0.06624563783407211,
      "learning_rate": 0.00019326960678900688,
      "loss": 10.2923,
      "step": 185
    },
    {
      "epoch": 0.02038398049274985,
      "grad_norm": 0.054792389273643494,
      "learning_rate": 0.00019319335541317361,
      "loss": 10.2936,
      "step": 186
    },
    {
      "epoch": 0.020493571785721625,
      "grad_norm": 0.05804865062236786,
      "learning_rate": 0.00019311668974014208,
      "loss": 10.2867,
      "step": 187
    },
    {
      "epoch": 0.020603163078693396,
      "grad_norm": 0.06261713802814484,
      "learning_rate": 0.00019303961011073447,
      "loss": 10.2879,
      "step": 188
    },
    {
      "epoch": 0.02071275437166517,
      "grad_norm": 0.05543063208460808,
      "learning_rate": 0.00019296211686761346,
      "loss": 10.2928,
      "step": 189
    },
    {
      "epoch": 0.020822345664636945,
      "grad_norm": 0.061838071793317795,
      "learning_rate": 0.00019288421035528028,
      "loss": 10.2878,
      "step": 190
    },
    {
      "epoch": 0.02093193695760872,
      "grad_norm": 0.04719708114862442,
      "learning_rate": 0.00019280589092007352,
      "loss": 10.2938,
      "step": 191
    },
    {
      "epoch": 0.02104152825058049,
      "grad_norm": 0.05846191197633743,
      "learning_rate": 0.00019272715891016735,
      "loss": 10.2847,
      "step": 192
    },
    {
      "epoch": 0.021151119543552265,
      "grad_norm": 0.05025499314069748,
      "learning_rate": 0.00019264801467557007,
      "loss": 10.2805,
      "step": 193
    },
    {
      "epoch": 0.02126071083652404,
      "grad_norm": 0.05473960563540459,
      "learning_rate": 0.00019256845856812266,
      "loss": 10.2929,
      "step": 194
    },
    {
      "epoch": 0.02137030212949581,
      "grad_norm": 0.057656463235616684,
      "learning_rate": 0.000192488490941497,
      "loss": 10.2886,
      "step": 195
    },
    {
      "epoch": 0.021479893422467585,
      "grad_norm": 0.060336221009492874,
      "learning_rate": 0.00019240811215119448,
      "loss": 10.2914,
      "step": 196
    },
    {
      "epoch": 0.02158948471543936,
      "grad_norm": 0.06550616770982742,
      "learning_rate": 0.00019232732255454422,
      "loss": 10.288,
      "step": 197
    },
    {
      "epoch": 0.02169907600841113,
      "grad_norm": 0.056043677031993866,
      "learning_rate": 0.00019224612251070175,
      "loss": 10.3059,
      "step": 198
    },
    {
      "epoch": 0.021808667301382905,
      "grad_norm": 0.07790745049715042,
      "learning_rate": 0.0001921645123806472,
      "loss": 10.2903,
      "step": 199
    },
    {
      "epoch": 0.02191825859435468,
      "grad_norm": 0.09033896774053574,
      "learning_rate": 0.0001920824925271838,
      "loss": 10.2924,
      "step": 200
    },
    {
      "epoch": 0.02202784988732645,
      "grad_norm": 0.06423371285200119,
      "learning_rate": 0.0001920000633149362,
      "loss": 10.2977,
      "step": 201
    },
    {
      "epoch": 0.022137441180298225,
      "grad_norm": 0.0567798838019371,
      "learning_rate": 0.00019191722511034884,
      "loss": 10.2958,
      "step": 202
    },
    {
      "epoch": 0.02224703247327,
      "grad_norm": 0.06479664891958237,
      "learning_rate": 0.00019183397828168448,
      "loss": 10.2954,
      "step": 203
    },
    {
      "epoch": 0.02235662376624177,
      "grad_norm": 0.04205462336540222,
      "learning_rate": 0.00019175032319902234,
      "loss": 10.2948,
      "step": 204
    },
    {
      "epoch": 0.022466215059213545,
      "grad_norm": 0.04938261955976486,
      "learning_rate": 0.00019166626023425662,
      "loss": 10.2971,
      "step": 205
    },
    {
      "epoch": 0.02257580635218532,
      "grad_norm": 0.05487078055739403,
      "learning_rate": 0.00019158178976109476,
      "loss": 10.2926,
      "step": 206
    },
    {
      "epoch": 0.022685397645157094,
      "grad_norm": 0.057584092020988464,
      "learning_rate": 0.0001914969121550558,
      "loss": 10.2961,
      "step": 207
    },
    {
      "epoch": 0.022794988938128865,
      "grad_norm": 0.05869193747639656,
      "learning_rate": 0.00019141162779346874,
      "loss": 10.2965,
      "step": 208
    },
    {
      "epoch": 0.02290458023110064,
      "grad_norm": 0.05281537026166916,
      "learning_rate": 0.00019132593705547082,
      "loss": 10.2951,
      "step": 209
    },
    {
      "epoch": 0.023014171524072414,
      "grad_norm": 0.056676436215639114,
      "learning_rate": 0.00019123984032200586,
      "loss": 10.2974,
      "step": 210
    },
    {
      "epoch": 0.023123762817044185,
      "grad_norm": 0.05642111599445343,
      "learning_rate": 0.00019115333797582254,
      "loss": 10.3072,
      "step": 211
    },
    {
      "epoch": 0.02323335411001596,
      "grad_norm": 0.0511922612786293,
      "learning_rate": 0.00019106643040147278,
      "loss": 10.301,
      "step": 212
    },
    {
      "epoch": 0.023342945402987734,
      "grad_norm": 0.05056412145495415,
      "learning_rate": 0.00019097911798530987,
      "loss": 10.2971,
      "step": 213
    },
    {
      "epoch": 0.023452536695959505,
      "grad_norm": 0.06952454894781113,
      "learning_rate": 0.00019089140111548696,
      "loss": 10.2949,
      "step": 214
    },
    {
      "epoch": 0.02356212798893128,
      "grad_norm": 0.056302182376384735,
      "learning_rate": 0.00019080328018195513,
      "loss": 10.2915,
      "step": 215
    },
    {
      "epoch": 0.023671719281903054,
      "grad_norm": 0.05767955631017685,
      "learning_rate": 0.0001907147555764618,
      "loss": 10.2944,
      "step": 216
    },
    {
      "epoch": 0.023781310574874825,
      "grad_norm": 0.03649438917636871,
      "learning_rate": 0.00019062582769254895,
      "loss": 10.2906,
      "step": 217
    },
    {
      "epoch": 0.0238909018678466,
      "grad_norm": 0.0575065091252327,
      "learning_rate": 0.00019053649692555135,
      "loss": 10.2926,
      "step": 218
    },
    {
      "epoch": 0.024000493160818374,
      "grad_norm": 0.03877263143658638,
      "learning_rate": 0.00019044676367259476,
      "loss": 10.293,
      "step": 219
    },
    {
      "epoch": 0.024110084453790145,
      "grad_norm": 0.046353355050086975,
      "learning_rate": 0.00019035662833259432,
      "loss": 10.2898,
      "step": 220
    },
    {
      "epoch": 0.02421967574676192,
      "grad_norm": 0.053095605224370956,
      "learning_rate": 0.00019026609130625257,
      "loss": 10.298,
      "step": 221
    },
    {
      "epoch": 0.024329267039733694,
      "grad_norm": 0.050848230719566345,
      "learning_rate": 0.00019017515299605788,
      "loss": 10.294,
      "step": 222
    },
    {
      "epoch": 0.024438858332705468,
      "grad_norm": 0.054096657782793045,
      "learning_rate": 0.00019008381380628247,
      "loss": 10.3012,
      "step": 223
    },
    {
      "epoch": 0.02454844962567724,
      "grad_norm": 0.06187242642045021,
      "learning_rate": 0.00018999207414298067,
      "loss": 10.291,
      "step": 224
    },
    {
      "epoch": 0.024658040918649014,
      "grad_norm": 0.05327076092362404,
      "learning_rate": 0.00018989993441398726,
      "loss": 10.2953,
      "step": 225
    },
    {
      "epoch": 0.024767632211620788,
      "grad_norm": 0.04818972945213318,
      "learning_rate": 0.00018980739502891546,
      "loss": 10.2975,
      "step": 226
    },
    {
      "epoch": 0.02487722350459256,
      "grad_norm": 0.053598254919052124,
      "learning_rate": 0.0001897144563991552,
      "loss": 10.2911,
      "step": 227
    },
    {
      "epoch": 0.024986814797564334,
      "grad_norm": 0.05477241426706314,
      "learning_rate": 0.00018962111893787128,
      "loss": 10.2957,
      "step": 228
    },
    {
      "epoch": 0.025096406090536108,
      "grad_norm": 0.06262899935245514,
      "learning_rate": 0.00018952738306000151,
      "loss": 10.2884,
      "step": 229
    },
    {
      "epoch": 0.02520599738350788,
      "grad_norm": 0.05376768857240677,
      "learning_rate": 0.00018943324918225494,
      "loss": 10.2899,
      "step": 230
    },
    {
      "epoch": 0.025315588676479654,
      "grad_norm": 0.058835215866565704,
      "learning_rate": 0.0001893387177231099,
      "loss": 10.2888,
      "step": 231
    },
    {
      "epoch": 0.025425179969451428,
      "grad_norm": 0.04260079190135002,
      "learning_rate": 0.0001892437891028122,
      "loss": 10.2898,
      "step": 232
    },
    {
      "epoch": 0.0255347712624232,
      "grad_norm": 0.055790942162275314,
      "learning_rate": 0.0001891484637433733,
      "loss": 10.2957,
      "step": 233
    },
    {
      "epoch": 0.025644362555394973,
      "grad_norm": 0.04617656022310257,
      "learning_rate": 0.00018905274206856837,
      "loss": 10.2948,
      "step": 234
    },
    {
      "epoch": 0.025753953848366748,
      "grad_norm": 0.044184610247612,
      "learning_rate": 0.00018895662450393438,
      "loss": 10.2841,
      "step": 235
    },
    {
      "epoch": 0.02586354514133852,
      "grad_norm": 0.06546135246753693,
      "learning_rate": 0.00018886011147676833,
      "loss": 10.2978,
      "step": 236
    },
    {
      "epoch": 0.025973136434310293,
      "grad_norm": 0.06937471777200699,
      "learning_rate": 0.00018876320341612522,
      "loss": 10.2958,
      "step": 237
    },
    {
      "epoch": 0.026082727727282068,
      "grad_norm": 0.048093002289533615,
      "learning_rate": 0.00018866590075281624,
      "loss": 10.2897,
      "step": 238
    },
    {
      "epoch": 0.026192319020253842,
      "grad_norm": 0.0589473620057106,
      "learning_rate": 0.00018856820391940674,
      "loss": 10.2923,
      "step": 239
    },
    {
      "epoch": 0.026301910313225613,
      "grad_norm": 0.07064120471477509,
      "learning_rate": 0.00018847011335021449,
      "loss": 10.295,
      "step": 240
    },
    {
      "epoch": 0.026411501606197388,
      "grad_norm": 0.06317900866270065,
      "learning_rate": 0.00018837162948130752,
      "loss": 10.2938,
      "step": 241
    },
    {
      "epoch": 0.026521092899169162,
      "grad_norm": 0.06477519869804382,
      "learning_rate": 0.00018827275275050233,
      "loss": 10.3005,
      "step": 242
    },
    {
      "epoch": 0.026630684192140933,
      "grad_norm": 0.038655828684568405,
      "learning_rate": 0.00018817348359736203,
      "loss": 10.2869,
      "step": 243
    },
    {
      "epoch": 0.026740275485112708,
      "grad_norm": 0.05626440420746803,
      "learning_rate": 0.00018807382246319412,
      "loss": 10.2939,
      "step": 244
    },
    {
      "epoch": 0.026849866778084482,
      "grad_norm": 0.052307456731796265,
      "learning_rate": 0.00018797376979104872,
      "loss": 10.2908,
      "step": 245
    },
    {
      "epoch": 0.026959458071056253,
      "grad_norm": 0.059653736650943756,
      "learning_rate": 0.00018787332602571662,
      "loss": 10.2892,
      "step": 246
    },
    {
      "epoch": 0.027069049364028028,
      "grad_norm": 0.06908773630857468,
      "learning_rate": 0.00018777249161372713,
      "loss": 10.2875,
      "step": 247
    },
    {
      "epoch": 0.027178640656999802,
      "grad_norm": 0.05643502250313759,
      "learning_rate": 0.00018767126700334634,
      "loss": 10.2964,
      "step": 248
    },
    {
      "epoch": 0.027288231949971573,
      "grad_norm": 0.06427492946386337,
      "learning_rate": 0.0001875696526445749,
      "loss": 10.3004,
      "step": 249
    },
    {
      "epoch": 0.027397823242943348,
      "grad_norm": 0.058874569833278656,
      "learning_rate": 0.0001874676489891461,
      "loss": 10.2953,
      "step": 250
    },
    {
      "epoch": 0.027507414535915122,
      "grad_norm": 0.049829546362161636,
      "learning_rate": 0.00018736525649052394,
      "loss": 10.302,
      "step": 251
    },
    {
      "epoch": 0.027617005828886897,
      "grad_norm": 0.049460217356681824,
      "learning_rate": 0.00018726247560390099,
      "loss": 10.2931,
      "step": 252
    },
    {
      "epoch": 0.027726597121858668,
      "grad_norm": 0.057572878897190094,
      "learning_rate": 0.00018715930678619644,
      "loss": 10.2949,
      "step": 253
    },
    {
      "epoch": 0.027836188414830442,
      "grad_norm": 0.0480433888733387,
      "learning_rate": 0.00018705575049605413,
      "loss": 10.297,
      "step": 254
    },
    {
      "epoch": 0.027945779707802217,
      "grad_norm": 0.050541676580905914,
      "learning_rate": 0.00018695180719384029,
      "loss": 10.2945,
      "step": 255
    },
    {
      "epoch": 0.028055371000773988,
      "grad_norm": 0.05799087509512901,
      "learning_rate": 0.00018684747734164177,
      "loss": 10.3001,
      "step": 256
    },
    {
      "epoch": 0.028164962293745762,
      "grad_norm": 0.04950912669301033,
      "learning_rate": 0.00018674276140326376,
      "loss": 10.297,
      "step": 257
    },
    {
      "epoch": 0.028274553586717537,
      "grad_norm": 0.05312972143292427,
      "learning_rate": 0.00018663765984422786,
      "loss": 10.3013,
      "step": 258
    },
    {
      "epoch": 0.028384144879689308,
      "grad_norm": 0.04490106180310249,
      "learning_rate": 0.00018653217313177004,
      "loss": 10.2943,
      "step": 259
    },
    {
      "epoch": 0.028493736172661082,
      "grad_norm": 0.04705547168850899,
      "learning_rate": 0.00018642630173483832,
      "loss": 10.2979,
      "step": 260
    },
    {
      "epoch": 0.028603327465632856,
      "grad_norm": 0.04983661696314812,
      "learning_rate": 0.00018632004612409103,
      "loss": 10.3034,
      "step": 261
    },
    {
      "epoch": 0.028712918758604628,
      "grad_norm": 0.06158896163105965,
      "learning_rate": 0.00018621340677189453,
      "loss": 10.2991,
      "step": 262
    },
    {
      "epoch": 0.028822510051576402,
      "grad_norm": 0.07297050207853317,
      "learning_rate": 0.00018610638415232097,
      "loss": 10.2968,
      "step": 263
    },
    {
      "epoch": 0.028932101344548176,
      "grad_norm": 0.0635574609041214,
      "learning_rate": 0.00018599897874114652,
      "loss": 10.2977,
      "step": 264
    },
    {
      "epoch": 0.029041692637519947,
      "grad_norm": 0.06697350740432739,
      "learning_rate": 0.00018589119101584898,
      "loss": 10.2969,
      "step": 265
    },
    {
      "epoch": 0.029151283930491722,
      "grad_norm": 0.0408085361123085,
      "learning_rate": 0.00018578302145560584,
      "loss": 10.2922,
      "step": 266
    },
    {
      "epoch": 0.029260875223463496,
      "grad_norm": 0.03781202435493469,
      "learning_rate": 0.00018567447054129195,
      "loss": 10.2953,
      "step": 267
    },
    {
      "epoch": 0.02937046651643527,
      "grad_norm": 0.050413526594638824,
      "learning_rate": 0.00018556553875547754,
      "loss": 10.3,
      "step": 268
    },
    {
      "epoch": 0.029480057809407042,
      "grad_norm": 0.041144102811813354,
      "learning_rate": 0.00018545622658242607,
      "loss": 10.2961,
      "step": 269
    },
    {
      "epoch": 0.029589649102378816,
      "grad_norm": 0.05554067716002464,
      "learning_rate": 0.00018534653450809197,
      "loss": 10.2924,
      "step": 270
    },
    {
      "epoch": 0.02969924039535059,
      "grad_norm": 0.04333749786019325,
      "learning_rate": 0.00018523646302011867,
      "loss": 10.3,
      "step": 271
    },
    {
      "epoch": 0.029808831688322362,
      "grad_norm": 0.04103844240307808,
      "learning_rate": 0.00018512601260783606,
      "loss": 10.2994,
      "step": 272
    },
    {
      "epoch": 0.029918422981294136,
      "grad_norm": 0.04418577998876572,
      "learning_rate": 0.00018501518376225887,
      "loss": 10.2862,
      "step": 273
    },
    {
      "epoch": 0.03002801427426591,
      "grad_norm": 0.05386722460389137,
      "learning_rate": 0.00018490397697608395,
      "loss": 10.2939,
      "step": 274
    },
    {
      "epoch": 0.030137605567237682,
      "grad_norm": 0.0518956296145916,
      "learning_rate": 0.0001847923927436884,
      "loss": 10.2928,
      "step": 275
    },
    {
      "epoch": 0.030247196860209456,
      "grad_norm": 0.054355405271053314,
      "learning_rate": 0.00018468043156112728,
      "loss": 10.2927,
      "step": 276
    },
    {
      "epoch": 0.03035678815318123,
      "grad_norm": 0.0494476854801178,
      "learning_rate": 0.0001845680939261314,
      "loss": 10.2929,
      "step": 277
    },
    {
      "epoch": 0.030466379446153,
      "grad_norm": 0.04498625919222832,
      "learning_rate": 0.00018445538033810515,
      "loss": 10.2914,
      "step": 278
    },
    {
      "epoch": 0.030575970739124776,
      "grad_norm": 0.05144556239247322,
      "learning_rate": 0.00018434229129812418,
      "loss": 10.2942,
      "step": 279
    },
    {
      "epoch": 0.03068556203209655,
      "grad_norm": 0.03634161874651909,
      "learning_rate": 0.0001842288273089332,
      "loss": 10.2909,
      "step": 280
    },
    {
      "epoch": 0.03079515332506832,
      "grad_norm": 0.04593217000365257,
      "learning_rate": 0.00018411498887494396,
      "loss": 10.2978,
      "step": 281
    },
    {
      "epoch": 0.030904744618040096,
      "grad_norm": 0.05329997092485428,
      "learning_rate": 0.00018400077650223263,
      "loss": 10.2877,
      "step": 282
    },
    {
      "epoch": 0.03101433591101187,
      "grad_norm": 0.05749444290995598,
      "learning_rate": 0.0001838861906985379,
      "loss": 10.2931,
      "step": 283
    },
    {
      "epoch": 0.031123927203983645,
      "grad_norm": 0.042937252670526505,
      "learning_rate": 0.00018377123197325842,
      "loss": 10.3017,
      "step": 284
    },
    {
      "epoch": 0.031233518496955416,
      "grad_norm": 0.05647299438714981,
      "learning_rate": 0.00018365590083745085,
      "loss": 10.2887,
      "step": 285
    },
    {
      "epoch": 0.03134310978992719,
      "grad_norm": 0.04040933772921562,
      "learning_rate": 0.00018354019780382735,
      "loss": 10.291,
      "step": 286
    },
    {
      "epoch": 0.031452701082898965,
      "grad_norm": 0.0491739921271801,
      "learning_rate": 0.0001834241233867533,
      "loss": 10.2878,
      "step": 287
    },
    {
      "epoch": 0.031562292375870736,
      "grad_norm": 0.04499319940805435,
      "learning_rate": 0.00018330767810224524,
      "loss": 10.2916,
      "step": 288
    },
    {
      "epoch": 0.03167188366884251,
      "grad_norm": 0.056227292865514755,
      "learning_rate": 0.0001831908624679683,
      "loss": 10.2932,
      "step": 289
    },
    {
      "epoch": 0.031781474961814285,
      "grad_norm": 0.045949049293994904,
      "learning_rate": 0.0001830736770032341,
      "loss": 10.291,
      "step": 290
    },
    {
      "epoch": 0.031891066254786056,
      "grad_norm": 0.04891704395413399,
      "learning_rate": 0.0001829561222289984,
      "loss": 10.2994,
      "step": 291
    },
    {
      "epoch": 0.032000657547757834,
      "grad_norm": 0.05303852632641792,
      "learning_rate": 0.00018283819866785853,
      "loss": 10.2853,
      "step": 292
    },
    {
      "epoch": 0.032110248840729605,
      "grad_norm": 0.0525033213198185,
      "learning_rate": 0.0001827199068440516,
      "loss": 10.2776,
      "step": 293
    },
    {
      "epoch": 0.032219840133701376,
      "grad_norm": 0.054567594081163406,
      "learning_rate": 0.00018260124728345162,
      "loss": 10.2874,
      "step": 294
    },
    {
      "epoch": 0.032329431426673154,
      "grad_norm": 0.0416351743042469,
      "learning_rate": 0.00018248222051356754,
      "loss": 10.2922,
      "step": 295
    },
    {
      "epoch": 0.032439022719644925,
      "grad_norm": 0.06239360570907593,
      "learning_rate": 0.00018236282706354063,
      "loss": 10.2952,
      "step": 296
    },
    {
      "epoch": 0.032548614012616696,
      "grad_norm": 0.07001176476478577,
      "learning_rate": 0.00018224306746414238,
      "loss": 10.2861,
      "step": 297
    },
    {
      "epoch": 0.032658205305588474,
      "grad_norm": 0.05840352550148964,
      "learning_rate": 0.00018212294224777197,
      "loss": 10.2847,
      "step": 298
    },
    {
      "epoch": 0.032767796598560245,
      "grad_norm": 0.06949873268604279,
      "learning_rate": 0.00018200245194845399,
      "loss": 10.296,
      "step": 299
    },
    {
      "epoch": 0.032877387891532016,
      "grad_norm": 0.08203252404928207,
      "learning_rate": 0.00018188159710183594,
      "loss": 10.2968,
      "step": 300
    },
    {
      "epoch": 0.032986979184503794,
      "grad_norm": 0.0552884042263031,
      "learning_rate": 0.000181760378245186,
      "loss": 10.2932,
      "step": 301
    },
    {
      "epoch": 0.033096570477475565,
      "grad_norm": 0.05044069141149521,
      "learning_rate": 0.00018163879591739067,
      "loss": 10.3001,
      "step": 302
    },
    {
      "epoch": 0.033206161770447336,
      "grad_norm": 0.06706534326076508,
      "learning_rate": 0.0001815168506589521,
      "loss": 10.2903,
      "step": 303
    },
    {
      "epoch": 0.033315753063419114,
      "grad_norm": 0.06277663260698318,
      "learning_rate": 0.000181394543011986,
      "loss": 10.2963,
      "step": 304
    },
    {
      "epoch": 0.033425344356390885,
      "grad_norm": 0.04979618266224861,
      "learning_rate": 0.00018127187352021907,
      "loss": 10.2988,
      "step": 305
    },
    {
      "epoch": 0.033534935649362656,
      "grad_norm": 0.04718731716275215,
      "learning_rate": 0.0001811488427289866,
      "loss": 10.2965,
      "step": 306
    },
    {
      "epoch": 0.033644526942334434,
      "grad_norm": 0.05341935530304909,
      "learning_rate": 0.00018102545118523007,
      "loss": 10.2965,
      "step": 307
    },
    {
      "epoch": 0.033754118235306205,
      "grad_norm": 0.056259654462337494,
      "learning_rate": 0.00018090169943749476,
      "loss": 10.2897,
      "step": 308
    },
    {
      "epoch": 0.033863709528277976,
      "grad_norm": 0.053433410823345184,
      "learning_rate": 0.00018077758803592718,
      "loss": 10.2904,
      "step": 309
    },
    {
      "epoch": 0.033973300821249754,
      "grad_norm": 0.039485760033130646,
      "learning_rate": 0.00018065311753227273,
      "loss": 10.295,
      "step": 310
    },
    {
      "epoch": 0.034082892114221525,
      "grad_norm": 0.045491062104701996,
      "learning_rate": 0.0001805282884798732,
      "loss": 10.297,
      "step": 311
    },
    {
      "epoch": 0.034192483407193296,
      "grad_norm": 0.051165759563446045,
      "learning_rate": 0.00018040310143366446,
      "loss": 10.3003,
      "step": 312
    },
    {
      "epoch": 0.034302074700165074,
      "grad_norm": 0.05436241999268532,
      "learning_rate": 0.00018027755695017368,
      "loss": 10.2951,
      "step": 313
    },
    {
      "epoch": 0.034411665993136845,
      "grad_norm": 0.06958349049091339,
      "learning_rate": 0.00018015165558751717,
      "loss": 10.29,
      "step": 314
    },
    {
      "epoch": 0.034521257286108616,
      "grad_norm": 0.07563730329275131,
      "learning_rate": 0.00018002539790539773,
      "loss": 10.295,
      "step": 315
    },
    {
      "epoch": 0.034630848579080394,
      "grad_norm": 0.04937342554330826,
      "learning_rate": 0.00017989878446510215,
      "loss": 10.2908,
      "step": 316
    },
    {
      "epoch": 0.034740439872052165,
      "grad_norm": 0.05015640705823898,
      "learning_rate": 0.00017977181582949888,
      "loss": 10.2904,
      "step": 317
    },
    {
      "epoch": 0.034850031165023936,
      "grad_norm": 0.05903152748942375,
      "learning_rate": 0.0001796444925630353,
      "loss": 10.2838,
      "step": 318
    },
    {
      "epoch": 0.034959622457995713,
      "grad_norm": 0.05885281041264534,
      "learning_rate": 0.00017951681523173542,
      "loss": 10.2919,
      "step": 319
    },
    {
      "epoch": 0.035069213750967484,
      "grad_norm": 0.05582178756594658,
      "learning_rate": 0.0001793887844031972,
      "loss": 10.2905,
      "step": 320
    },
    {
      "epoch": 0.03517880504393926,
      "grad_norm": 0.05661908909678459,
      "learning_rate": 0.00017926040064659014,
      "loss": 10.2952,
      "step": 321
    },
    {
      "epoch": 0.03528839633691103,
      "grad_norm": 0.048182882368564606,
      "learning_rate": 0.0001791316645326526,
      "loss": 10.2921,
      "step": 322
    },
    {
      "epoch": 0.035397987629882804,
      "grad_norm": 0.05921144410967827,
      "learning_rate": 0.00017900257663368963,
      "loss": 10.2853,
      "step": 323
    },
    {
      "epoch": 0.03550757892285458,
      "grad_norm": 0.051330771297216415,
      "learning_rate": 0.0001788731375235698,
      "loss": 10.2904,
      "step": 324
    },
    {
      "epoch": 0.03561717021582635,
      "grad_norm": 0.05465542525053024,
      "learning_rate": 0.00017874334777772327,
      "loss": 10.29,
      "step": 325
    },
    {
      "epoch": 0.035726761508798124,
      "grad_norm": 0.05385803431272507,
      "learning_rate": 0.00017861320797313892,
      "loss": 10.2904,
      "step": 326
    },
    {
      "epoch": 0.0358363528017699,
      "grad_norm": 0.04716626554727554,
      "learning_rate": 0.0001784827186883618,
      "loss": 10.2936,
      "step": 327
    },
    {
      "epoch": 0.03594594409474167,
      "grad_norm": 0.033427998423576355,
      "learning_rate": 0.00017835188050349064,
      "loss": 10.2929,
      "step": 328
    },
    {
      "epoch": 0.036055535387713444,
      "grad_norm": 0.047436170279979706,
      "learning_rate": 0.00017822069400017516,
      "loss": 10.2943,
      "step": 329
    },
    {
      "epoch": 0.03616512668068522,
      "grad_norm": 0.046525925397872925,
      "learning_rate": 0.00017808915976161362,
      "loss": 10.2804,
      "step": 330
    },
    {
      "epoch": 0.03627471797365699,
      "grad_norm": 0.04504426196217537,
      "learning_rate": 0.00017795727837255015,
      "loss": 10.2972,
      "step": 331
    },
    {
      "epoch": 0.036384309266628764,
      "grad_norm": 0.05005035549402237,
      "learning_rate": 0.00017782505041927216,
      "loss": 10.2942,
      "step": 332
    },
    {
      "epoch": 0.03649390055960054,
      "grad_norm": 0.03621133789420128,
      "learning_rate": 0.00017769247648960774,
      "loss": 10.2929,
      "step": 333
    },
    {
      "epoch": 0.03660349185257231,
      "grad_norm": 0.051758553832769394,
      "learning_rate": 0.00017755955717292296,
      "loss": 10.288,
      "step": 334
    },
    {
      "epoch": 0.036713083145544084,
      "grad_norm": 0.053045254200696945,
      "learning_rate": 0.00017742629306011944,
      "loss": 10.2872,
      "step": 335
    },
    {
      "epoch": 0.03682267443851586,
      "grad_norm": 0.045028284192085266,
      "learning_rate": 0.00017729268474363154,
      "loss": 10.2881,
      "step": 336
    },
    {
      "epoch": 0.03693226573148763,
      "grad_norm": 0.053224459290504456,
      "learning_rate": 0.0001771587328174239,
      "loss": 10.2787,
      "step": 337
    },
    {
      "epoch": 0.037041857024459404,
      "grad_norm": 0.05798415467143059,
      "learning_rate": 0.0001770244378769885,
      "loss": 10.2866,
      "step": 338
    },
    {
      "epoch": 0.03715144831743118,
      "grad_norm": 0.06549213826656342,
      "learning_rate": 0.0001768898005193425,
      "loss": 10.2863,
      "step": 339
    },
    {
      "epoch": 0.03726103961040295,
      "grad_norm": 0.06039006635546684,
      "learning_rate": 0.000176754821343025,
      "loss": 10.2974,
      "step": 340
    },
    {
      "epoch": 0.037370630903374724,
      "grad_norm": 0.05214457958936691,
      "learning_rate": 0.0001766195009480949,
      "loss": 10.295,
      "step": 341
    },
    {
      "epoch": 0.0374802221963465,
      "grad_norm": 0.06046038120985031,
      "learning_rate": 0.0001764838399361279,
      "loss": 10.2865,
      "step": 342
    },
    {
      "epoch": 0.03758981348931827,
      "grad_norm": 0.05095124617218971,
      "learning_rate": 0.00017634783891021393,
      "loss": 10.2912,
      "step": 343
    },
    {
      "epoch": 0.037699404782290044,
      "grad_norm": 0.055187124758958817,
      "learning_rate": 0.00017621149847495458,
      "loss": 10.2904,
      "step": 344
    },
    {
      "epoch": 0.03780899607526182,
      "grad_norm": 0.06513886898756027,
      "learning_rate": 0.00017607481923646016,
      "loss": 10.2883,
      "step": 345
    },
    {
      "epoch": 0.03791858736823359,
      "grad_norm": 0.058212023228406906,
      "learning_rate": 0.0001759378018023473,
      "loss": 10.2957,
      "step": 346
    },
    {
      "epoch": 0.038028178661205364,
      "grad_norm": 0.05906583368778229,
      "learning_rate": 0.00017580044678173592,
      "loss": 10.2977,
      "step": 347
    },
    {
      "epoch": 0.03813776995417714,
      "grad_norm": 0.0634850263595581,
      "learning_rate": 0.00017566275478524693,
      "loss": 10.2921,
      "step": 348
    },
    {
      "epoch": 0.03824736124714891,
      "grad_norm": 0.06109980121254921,
      "learning_rate": 0.0001755247264249991,
      "loss": 10.2825,
      "step": 349
    },
    {
      "epoch": 0.038356952540120684,
      "grad_norm": 0.05886857956647873,
      "learning_rate": 0.0001753863623146066,
      "loss": 10.2935,
      "step": 350
    },
    {
      "epoch": 0.03846654383309246,
      "grad_norm": 0.05367111414670944,
      "learning_rate": 0.00017524766306917618,
      "loss": 10.2972,
      "step": 351
    },
    {
      "epoch": 0.03857613512606423,
      "grad_norm": 0.054326120764017105,
      "learning_rate": 0.0001751086293053045,
      "loss": 10.29,
      "step": 352
    },
    {
      "epoch": 0.03868572641903601,
      "grad_norm": 0.0628490149974823,
      "learning_rate": 0.0001749692616410753,
      "loss": 10.2987,
      "step": 353
    },
    {
      "epoch": 0.03879531771200778,
      "grad_norm": 0.0647919550538063,
      "learning_rate": 0.00017482956069605668,
      "loss": 10.295,
      "step": 354
    },
    {
      "epoch": 0.03890490900497955,
      "grad_norm": 0.05330973118543625,
      "learning_rate": 0.00017468952709129846,
      "loss": 10.2905,
      "step": 355
    },
    {
      "epoch": 0.03901450029795133,
      "grad_norm": 0.05702512338757515,
      "learning_rate": 0.00017454916144932922,
      "loss": 10.2913,
      "step": 356
    },
    {
      "epoch": 0.0391240915909231,
      "grad_norm": 0.059527911245822906,
      "learning_rate": 0.0001744084643941536,
      "loss": 10.293,
      "step": 357
    },
    {
      "epoch": 0.03923368288389487,
      "grad_norm": 0.05024813860654831,
      "learning_rate": 0.00017426743655124974,
      "loss": 10.297,
      "step": 358
    },
    {
      "epoch": 0.03934327417686665,
      "grad_norm": 0.05896433815360069,
      "learning_rate": 0.0001741260785475661,
      "loss": 10.2987,
      "step": 359
    },
    {
      "epoch": 0.03945286546983842,
      "grad_norm": 0.04255250096321106,
      "learning_rate": 0.00017398439101151905,
      "loss": 10.2923,
      "step": 360
    },
    {
      "epoch": 0.03956245676281019,
      "grad_norm": 0.047444187104701996,
      "learning_rate": 0.00017384237457298987,
      "loss": 10.3015,
      "step": 361
    },
    {
      "epoch": 0.03967204805578197,
      "grad_norm": 0.05658761039376259,
      "learning_rate": 0.00017370002986332193,
      "loss": 10.289,
      "step": 362
    },
    {
      "epoch": 0.03978163934875374,
      "grad_norm": 0.05257973447442055,
      "learning_rate": 0.00017355735751531807,
      "loss": 10.2923,
      "step": 363
    },
    {
      "epoch": 0.03989123064172551,
      "grad_norm": 0.059355076402425766,
      "learning_rate": 0.00017341435816323756,
      "loss": 10.293,
      "step": 364
    },
    {
      "epoch": 0.04000082193469729,
      "grad_norm": 0.05914102494716644,
      "learning_rate": 0.00017327103244279348,
      "loss": 10.2947,
      "step": 365
    },
    {
      "epoch": 0.04011041322766906,
      "grad_norm": 0.056688327342271805,
      "learning_rate": 0.00017312738099114973,
      "loss": 10.2859,
      "step": 366
    },
    {
      "epoch": 0.04022000452064083,
      "grad_norm": 0.048409488052129745,
      "learning_rate": 0.00017298340444691835,
      "loss": 10.2883,
      "step": 367
    },
    {
      "epoch": 0.04032959581361261,
      "grad_norm": 0.04540451988577843,
      "learning_rate": 0.00017283910345015647,
      "loss": 10.2946,
      "step": 368
    },
    {
      "epoch": 0.04043918710658438,
      "grad_norm": 0.04332428798079491,
      "learning_rate": 0.0001726944786423637,
      "loss": 10.2845,
      "step": 369
    },
    {
      "epoch": 0.04054877839955615,
      "grad_norm": 0.043001867830753326,
      "learning_rate": 0.00017254953066647913,
      "loss": 10.2897,
      "step": 370
    },
    {
      "epoch": 0.04065836969252793,
      "grad_norm": 0.04882137104868889,
      "learning_rate": 0.00017240426016687863,
      "loss": 10.2867,
      "step": 371
    },
    {
      "epoch": 0.0407679609854997,
      "grad_norm": 0.05311378464102745,
      "learning_rate": 0.00017225866778937165,
      "loss": 10.2907,
      "step": 372
    },
    {
      "epoch": 0.04087755227847147,
      "grad_norm": 0.06117072328925133,
      "learning_rate": 0.00017211275418119876,
      "loss": 10.2828,
      "step": 373
    },
    {
      "epoch": 0.04098714357144325,
      "grad_norm": 0.03560377284884453,
      "learning_rate": 0.0001719665199910285,
      "loss": 10.2931,
      "step": 374
    },
    {
      "epoch": 0.04109673486441502,
      "grad_norm": 0.0595746785402298,
      "learning_rate": 0.00017181996586895454,
      "loss": 10.293,
      "step": 375
    },
    {
      "epoch": 0.04109673486441502,
      "eval_loss": 10.288755416870117,
      "eval_runtime": 54.9967,
      "eval_samples_per_second": 279.453,
      "eval_steps_per_second": 139.736,
      "step": 375
    },
    {
      "epoch": 0.04120632615738679,
      "grad_norm": 0.04847024753689766,
      "learning_rate": 0.00017167309246649297,
      "loss": 10.293,
      "step": 376
    },
    {
      "epoch": 0.04131591745035857,
      "grad_norm": 0.06213182583451271,
      "learning_rate": 0.0001715259004365791,
      "loss": 10.2919,
      "step": 377
    },
    {
      "epoch": 0.04142550874333034,
      "grad_norm": 0.05493413656949997,
      "learning_rate": 0.00017137839043356484,
      "loss": 10.2962,
      "step": 378
    },
    {
      "epoch": 0.04153510003630211,
      "grad_norm": 0.05115830898284912,
      "learning_rate": 0.00017123056311321562,
      "loss": 10.2853,
      "step": 379
    },
    {
      "epoch": 0.04164469132927389,
      "grad_norm": 0.06038088724017143,
      "learning_rate": 0.0001710824191327075,
      "loss": 10.275,
      "step": 380
    },
    {
      "epoch": 0.04175428262224566,
      "grad_norm": 0.05692605674266815,
      "learning_rate": 0.00017093395915062428,
      "loss": 10.2865,
      "step": 381
    },
    {
      "epoch": 0.04186387391521744,
      "grad_norm": 0.04580005630850792,
      "learning_rate": 0.00017078518382695465,
      "loss": 10.2988,
      "step": 382
    },
    {
      "epoch": 0.04197346520818921,
      "grad_norm": 0.05712489038705826,
      "learning_rate": 0.00017063609382308908,
      "loss": 10.2816,
      "step": 383
    },
    {
      "epoch": 0.04208305650116098,
      "grad_norm": 0.063704714179039,
      "learning_rate": 0.00017048668980181698,
      "loss": 10.2769,
      "step": 384
    },
    {
      "epoch": 0.04219264779413276,
      "grad_norm": 0.05224073305726051,
      "learning_rate": 0.00017033697242732377,
      "loss": 10.286,
      "step": 385
    },
    {
      "epoch": 0.04230223908710453,
      "grad_norm": 0.07389599084854126,
      "learning_rate": 0.0001701869423651879,
      "loss": 10.288,
      "step": 386
    },
    {
      "epoch": 0.0424118303800763,
      "grad_norm": 0.05292465537786484,
      "learning_rate": 0.00017003660028237793,
      "loss": 10.2826,
      "step": 387
    },
    {
      "epoch": 0.04252142167304808,
      "grad_norm": 0.06110900640487671,
      "learning_rate": 0.00016988594684724947,
      "loss": 10.2769,
      "step": 388
    },
    {
      "epoch": 0.04263101296601985,
      "grad_norm": 0.056761983782052994,
      "learning_rate": 0.00016973498272954222,
      "loss": 10.2864,
      "step": 389
    },
    {
      "epoch": 0.04274060425899162,
      "grad_norm": 0.04813006520271301,
      "learning_rate": 0.00016958370860037717,
      "loss": 10.2908,
      "step": 390
    },
    {
      "epoch": 0.0428501955519634,
      "grad_norm": 0.04608007147908211,
      "learning_rate": 0.00016943212513225345,
      "loss": 10.2858,
      "step": 391
    },
    {
      "epoch": 0.04295978684493517,
      "grad_norm": 0.05717607960104942,
      "learning_rate": 0.00016928023299904533,
      "loss": 10.2787,
      "step": 392
    },
    {
      "epoch": 0.04306937813790694,
      "grad_norm": 0.05666106939315796,
      "learning_rate": 0.0001691280328759992,
      "loss": 10.2793,
      "step": 393
    },
    {
      "epoch": 0.04317896943087872,
      "grad_norm": 0.06093747541308403,
      "learning_rate": 0.00016897552543973084,
      "loss": 10.2858,
      "step": 394
    },
    {
      "epoch": 0.04328856072385049,
      "grad_norm": 0.06447304785251617,
      "learning_rate": 0.00016882271136822206,
      "loss": 10.2844,
      "step": 395
    },
    {
      "epoch": 0.04339815201682226,
      "grad_norm": 0.06769227981567383,
      "learning_rate": 0.0001686695913408179,
      "loss": 10.2943,
      "step": 396
    },
    {
      "epoch": 0.04350774330979404,
      "grad_norm": 0.06110338866710663,
      "learning_rate": 0.0001685161660382235,
      "loss": 10.2847,
      "step": 397
    },
    {
      "epoch": 0.04361733460276581,
      "grad_norm": 0.053588930517435074,
      "learning_rate": 0.00016836243614250113,
      "loss": 10.2781,
      "step": 398
    },
    {
      "epoch": 0.04372692589573758,
      "grad_norm": 0.07180412858724594,
      "learning_rate": 0.00016820840233706719,
      "loss": 10.284,
      "step": 399
    },
    {
      "epoch": 0.04383651718870936,
      "grad_norm": 0.07615606486797333,
      "learning_rate": 0.0001680540653066891,
      "loss": 10.2901,
      "step": 400
    },
    {
      "epoch": 0.04394610848168113,
      "grad_norm": 0.04890253767371178,
      "learning_rate": 0.00016789942573748232,
      "loss": 10.2946,
      "step": 401
    },
    {
      "epoch": 0.0440556997746529,
      "grad_norm": 0.046120256185531616,
      "learning_rate": 0.0001677444843169072,
      "loss": 10.2936,
      "step": 402
    },
    {
      "epoch": 0.04416529106762468,
      "grad_norm": 0.05320224538445473,
      "learning_rate": 0.00016758924173376603,
      "loss": 10.2925,
      "step": 403
    },
    {
      "epoch": 0.04427488236059645,
      "grad_norm": 0.05933986231684685,
      "learning_rate": 0.0001674336986781999,
      "loss": 10.291,
      "step": 404
    },
    {
      "epoch": 0.04438447365356822,
      "grad_norm": 0.04727527126669884,
      "learning_rate": 0.00016727785584168581,
      "loss": 10.2938,
      "step": 405
    },
    {
      "epoch": 0.04449406494654,
      "grad_norm": 0.05204232037067413,
      "learning_rate": 0.0001671217139170333,
      "loss": 10.2897,
      "step": 406
    },
    {
      "epoch": 0.04460365623951177,
      "grad_norm": 0.04568104445934296,
      "learning_rate": 0.00016696527359838154,
      "loss": 10.2877,
      "step": 407
    },
    {
      "epoch": 0.04471324753248354,
      "grad_norm": 0.05282888934016228,
      "learning_rate": 0.00016680853558119632,
      "loss": 10.288,
      "step": 408
    },
    {
      "epoch": 0.04482283882545532,
      "grad_norm": 0.052875179797410965,
      "learning_rate": 0.0001666515005622668,
      "loss": 10.2857,
      "step": 409
    },
    {
      "epoch": 0.04493243011842709,
      "grad_norm": 0.05460058152675629,
      "learning_rate": 0.0001664941692397025,
      "loss": 10.2783,
      "step": 410
    },
    {
      "epoch": 0.04504202141139886,
      "grad_norm": 0.04738298058509827,
      "learning_rate": 0.00016633654231293013,
      "loss": 10.3002,
      "step": 411
    },
    {
      "epoch": 0.04515161270437064,
      "grad_norm": 0.06071951240301132,
      "learning_rate": 0.00016617862048269065,
      "loss": 10.2897,
      "step": 412
    },
    {
      "epoch": 0.04526120399734241,
      "grad_norm": 0.056846972554922104,
      "learning_rate": 0.00016602040445103588,
      "loss": 10.2869,
      "step": 413
    },
    {
      "epoch": 0.04537079529031419,
      "grad_norm": 0.04261229187250137,
      "learning_rate": 0.00016586189492132566,
      "loss": 10.2941,
      "step": 414
    },
    {
      "epoch": 0.04548038658328596,
      "grad_norm": 0.05167244002223015,
      "learning_rate": 0.00016570309259822453,
      "loss": 10.2953,
      "step": 415
    },
    {
      "epoch": 0.04558997787625773,
      "grad_norm": 0.04831593856215477,
      "learning_rate": 0.0001655439981876987,
      "loss": 10.2909,
      "step": 416
    },
    {
      "epoch": 0.04569956916922951,
      "grad_norm": 0.055484164506196976,
      "learning_rate": 0.00016538461239701277,
      "loss": 10.2913,
      "step": 417
    },
    {
      "epoch": 0.04580916046220128,
      "grad_norm": 0.04777919501066208,
      "learning_rate": 0.00016522493593472683,
      "loss": 10.2826,
      "step": 418
    },
    {
      "epoch": 0.04591875175517305,
      "grad_norm": 0.050060734152793884,
      "learning_rate": 0.0001650649695106931,
      "loss": 10.2836,
      "step": 419
    },
    {
      "epoch": 0.04602834304814483,
      "grad_norm": 0.045418158173561096,
      "learning_rate": 0.00016490471383605288,
      "loss": 10.2887,
      "step": 420
    },
    {
      "epoch": 0.0461379343411166,
      "grad_norm": 0.051229026168584824,
      "learning_rate": 0.00016474416962323325,
      "loss": 10.2898,
      "step": 421
    },
    {
      "epoch": 0.04624752563408837,
      "grad_norm": 0.0563134029507637,
      "learning_rate": 0.00016458333758594414,
      "loss": 10.2877,
      "step": 422
    },
    {
      "epoch": 0.04635711692706015,
      "grad_norm": 0.04452783241868019,
      "learning_rate": 0.00016442221843917496,
      "loss": 10.2841,
      "step": 423
    },
    {
      "epoch": 0.04646670822003192,
      "grad_norm": 0.06369493156671524,
      "learning_rate": 0.00016426081289919143,
      "loss": 10.2894,
      "step": 424
    },
    {
      "epoch": 0.04657629951300369,
      "grad_norm": 0.06141071766614914,
      "learning_rate": 0.0001640991216835326,
      "loss": 10.2908,
      "step": 425
    },
    {
      "epoch": 0.04668589080597547,
      "grad_norm": 0.06164323166012764,
      "learning_rate": 0.00016393714551100734,
      "loss": 10.2788,
      "step": 426
    },
    {
      "epoch": 0.04679548209894724,
      "grad_norm": 0.054539553821086884,
      "learning_rate": 0.0001637748851016914,
      "loss": 10.2934,
      "step": 427
    },
    {
      "epoch": 0.04690507339191901,
      "grad_norm": 0.05947967618703842,
      "learning_rate": 0.00016361234117692413,
      "loss": 10.2852,
      "step": 428
    },
    {
      "epoch": 0.04701466468489079,
      "grad_norm": 0.05250566080212593,
      "learning_rate": 0.00016344951445930526,
      "loss": 10.2785,
      "step": 429
    },
    {
      "epoch": 0.04712425597786256,
      "grad_norm": 0.050202399492263794,
      "learning_rate": 0.0001632864056726917,
      "loss": 10.284,
      "step": 430
    },
    {
      "epoch": 0.04723384727083433,
      "grad_norm": 0.044646598398685455,
      "learning_rate": 0.00016312301554219426,
      "loss": 10.2835,
      "step": 431
    },
    {
      "epoch": 0.04734343856380611,
      "grad_norm": 0.05671941488981247,
      "learning_rate": 0.00016295934479417453,
      "loss": 10.2827,
      "step": 432
    },
    {
      "epoch": 0.04745302985677788,
      "grad_norm": 0.05306157469749451,
      "learning_rate": 0.00016279539415624164,
      "loss": 10.2911,
      "step": 433
    },
    {
      "epoch": 0.04756262114974965,
      "grad_norm": 0.05661826580762863,
      "learning_rate": 0.0001626311643572489,
      "loss": 10.2798,
      "step": 434
    },
    {
      "epoch": 0.04767221244272143,
      "grad_norm": 0.059121664613485336,
      "learning_rate": 0.00016246665612729074,
      "loss": 10.2834,
      "step": 435
    },
    {
      "epoch": 0.0477818037356932,
      "grad_norm": 0.06919396668672562,
      "learning_rate": 0.00016230187019769928,
      "loss": 10.2793,
      "step": 436
    },
    {
      "epoch": 0.04789139502866497,
      "grad_norm": 0.05846366658806801,
      "learning_rate": 0.00016213680730104124,
      "loss": 10.2846,
      "step": 437
    },
    {
      "epoch": 0.04800098632163675,
      "grad_norm": 0.06126514822244644,
      "learning_rate": 0.0001619714681711146,
      "loss": 10.2762,
      "step": 438
    },
    {
      "epoch": 0.04811057761460852,
      "grad_norm": 0.0590432770550251,
      "learning_rate": 0.00016180585354294536,
      "loss": 10.276,
      "step": 439
    },
    {
      "epoch": 0.04822016890758029,
      "grad_norm": 0.06297392398118973,
      "learning_rate": 0.00016163996415278424,
      "loss": 10.2789,
      "step": 440
    },
    {
      "epoch": 0.04832976020055207,
      "grad_norm": 0.06185074895620346,
      "learning_rate": 0.00016147380073810346,
      "loss": 10.2732,
      "step": 441
    },
    {
      "epoch": 0.04843935149352384,
      "grad_norm": 0.055530861020088196,
      "learning_rate": 0.0001613073640375934,
      "loss": 10.2886,
      "step": 442
    },
    {
      "epoch": 0.048548942786495616,
      "grad_norm": 0.05897903814911842,
      "learning_rate": 0.00016114065479115946,
      "loss": 10.2887,
      "step": 443
    },
    {
      "epoch": 0.04865853407946739,
      "grad_norm": 0.05224940925836563,
      "learning_rate": 0.00016097367373991842,
      "loss": 10.2836,
      "step": 444
    },
    {
      "epoch": 0.04876812537243916,
      "grad_norm": 0.06540297716856003,
      "learning_rate": 0.00016080642162619565,
      "loss": 10.278,
      "step": 445
    },
    {
      "epoch": 0.048877716665410936,
      "grad_norm": 0.06439578533172607,
      "learning_rate": 0.0001606388991935214,
      "loss": 10.283,
      "step": 446
    },
    {
      "epoch": 0.04898730795838271,
      "grad_norm": 0.05347806215286255,
      "learning_rate": 0.0001604711071866277,
      "loss": 10.2861,
      "step": 447
    },
    {
      "epoch": 0.04909689925135448,
      "grad_norm": 0.06549952924251556,
      "learning_rate": 0.00016030304635144494,
      "loss": 10.2808,
      "step": 448
    },
    {
      "epoch": 0.049206490544326256,
      "grad_norm": 0.06713661551475525,
      "learning_rate": 0.00016013471743509862,
      "loss": 10.2847,
      "step": 449
    },
    {
      "epoch": 0.04931608183729803,
      "grad_norm": 0.07478884607553482,
      "learning_rate": 0.00015996612118590603,
      "loss": 10.2854,
      "step": 450
    },
    {
      "epoch": 0.0494256731302698,
      "grad_norm": 0.04559367150068283,
      "learning_rate": 0.00015979725835337294,
      "loss": 10.2907,
      "step": 451
    },
    {
      "epoch": 0.049535264423241576,
      "grad_norm": 0.047304823994636536,
      "learning_rate": 0.00015962812968819016,
      "loss": 10.2877,
      "step": 452
    },
    {
      "epoch": 0.04964485571621335,
      "grad_norm": 0.047875773161649704,
      "learning_rate": 0.0001594587359422303,
      "loss": 10.2845,
      "step": 453
    },
    {
      "epoch": 0.04975444700918512,
      "grad_norm": 0.044304296374320984,
      "learning_rate": 0.0001592890778685444,
      "loss": 10.2851,
      "step": 454
    },
    {
      "epoch": 0.049864038302156896,
      "grad_norm": 0.05559462681412697,
      "learning_rate": 0.00015911915622135862,
      "loss": 10.2807,
      "step": 455
    },
    {
      "epoch": 0.04997362959512867,
      "grad_norm": 0.05317340046167374,
      "learning_rate": 0.00015894897175607086,
      "loss": 10.2818,
      "step": 456
    },
    {
      "epoch": 0.05008322088810044,
      "grad_norm": 0.054345984011888504,
      "learning_rate": 0.00015877852522924732,
      "loss": 10.2906,
      "step": 457
    },
    {
      "epoch": 0.050192812181072216,
      "grad_norm": 0.05537887662649155,
      "learning_rate": 0.00015860781739861928,
      "loss": 10.2916,
      "step": 458
    },
    {
      "epoch": 0.05030240347404399,
      "grad_norm": 0.06797152012586594,
      "learning_rate": 0.00015843684902307962,
      "loss": 10.3006,
      "step": 459
    },
    {
      "epoch": 0.05041199476701576,
      "grad_norm": 0.0537358857691288,
      "learning_rate": 0.00015826562086267956,
      "loss": 10.2895,
      "step": 460
    },
    {
      "epoch": 0.050521586059987536,
      "grad_norm": 0.050809480249881744,
      "learning_rate": 0.00015809413367862512,
      "loss": 10.2872,
      "step": 461
    },
    {
      "epoch": 0.05063117735295931,
      "grad_norm": 0.06265373528003693,
      "learning_rate": 0.00015792238823327388,
      "loss": 10.2945,
      "step": 462
    },
    {
      "epoch": 0.05074076864593108,
      "grad_norm": 0.06667222827672958,
      "learning_rate": 0.00015775038529013152,
      "loss": 10.289,
      "step": 463
    },
    {
      "epoch": 0.050850359938902856,
      "grad_norm": 0.05941317230463028,
      "learning_rate": 0.0001575781256138485,
      "loss": 10.2943,
      "step": 464
    },
    {
      "epoch": 0.05095995123187463,
      "grad_norm": 0.07160047441720963,
      "learning_rate": 0.00015740560997021648,
      "loss": 10.2913,
      "step": 465
    },
    {
      "epoch": 0.0510695425248464,
      "grad_norm": 0.050766777247190475,
      "learning_rate": 0.00015723283912616513,
      "loss": 10.2939,
      "step": 466
    },
    {
      "epoch": 0.051179133817818176,
      "grad_norm": 0.04814283922314644,
      "learning_rate": 0.00015705981384975866,
      "loss": 10.2845,
      "step": 467
    },
    {
      "epoch": 0.05128872511078995,
      "grad_norm": 0.05674029141664505,
      "learning_rate": 0.0001568865349101923,
      "loss": 10.2826,
      "step": 468
    },
    {
      "epoch": 0.05139831640376172,
      "grad_norm": 0.041860077530145645,
      "learning_rate": 0.00015671300307778898,
      "loss": 10.281,
      "step": 469
    },
    {
      "epoch": 0.051507907696733496,
      "grad_norm": 0.0534890815615654,
      "learning_rate": 0.00015653921912399589,
      "loss": 10.276,
      "step": 470
    },
    {
      "epoch": 0.05161749898970527,
      "grad_norm": 0.044509273022413254,
      "learning_rate": 0.00015636518382138107,
      "loss": 10.2852,
      "step": 471
    },
    {
      "epoch": 0.05172709028267704,
      "grad_norm": 0.05128435045480728,
      "learning_rate": 0.0001561908979436299,
      "loss": 10.2831,
      "step": 472
    },
    {
      "epoch": 0.051836681575648816,
      "grad_norm": 0.05868786945939064,
      "learning_rate": 0.00015601636226554168,
      "loss": 10.2866,
      "step": 473
    },
    {
      "epoch": 0.05194627286862059,
      "grad_norm": 0.053306397050619125,
      "learning_rate": 0.00015584157756302634,
      "loss": 10.2874,
      "step": 474
    },
    {
      "epoch": 0.052055864161592365,
      "grad_norm": 0.043936919420957565,
      "learning_rate": 0.0001556665446131007,
      "loss": 10.282,
      "step": 475
    },
    {
      "epoch": 0.052165455454564136,
      "grad_norm": 0.05355142802000046,
      "learning_rate": 0.00015549126419388536,
      "loss": 10.2773,
      "step": 476
    },
    {
      "epoch": 0.05227504674753591,
      "grad_norm": 0.051142219454050064,
      "learning_rate": 0.0001553157370846009,
      "loss": 10.2842,
      "step": 477
    },
    {
      "epoch": 0.052384638040507685,
      "grad_norm": 0.051057569682598114,
      "learning_rate": 0.00015513996406556465,
      "loss": 10.2801,
      "step": 478
    },
    {
      "epoch": 0.052494229333479456,
      "grad_norm": 0.047144535928964615,
      "learning_rate": 0.00015496394591818716,
      "loss": 10.2862,
      "step": 479
    },
    {
      "epoch": 0.05260382062645123,
      "grad_norm": 0.052774276584386826,
      "learning_rate": 0.0001547876834249687,
      "loss": 10.2864,
      "step": 480
    },
    {
      "epoch": 0.052713411919423005,
      "grad_norm": 0.05825953558087349,
      "learning_rate": 0.00015461117736949577,
      "loss": 10.27,
      "step": 481
    },
    {
      "epoch": 0.052823003212394776,
      "grad_norm": 0.05014825984835625,
      "learning_rate": 0.00015443442853643762,
      "loss": 10.2809,
      "step": 482
    },
    {
      "epoch": 0.05293259450536655,
      "grad_norm": 0.04590538144111633,
      "learning_rate": 0.00015425743771154294,
      "loss": 10.2799,
      "step": 483
    },
    {
      "epoch": 0.053042185798338325,
      "grad_norm": 0.05817478150129318,
      "learning_rate": 0.00015408020568163602,
      "loss": 10.2761,
      "step": 484
    },
    {
      "epoch": 0.053151777091310096,
      "grad_norm": 0.05610961467027664,
      "learning_rate": 0.00015390273323461352,
      "loss": 10.2802,
      "step": 485
    },
    {
      "epoch": 0.05326136838428187,
      "grad_norm": 0.05526053532958031,
      "learning_rate": 0.0001537250211594409,
      "loss": 10.2929,
      "step": 486
    },
    {
      "epoch": 0.053370959677253645,
      "grad_norm": 0.07304730266332626,
      "learning_rate": 0.0001535470702461489,
      "loss": 10.2749,
      "step": 487
    },
    {
      "epoch": 0.053480550970225416,
      "grad_norm": 0.057518765330314636,
      "learning_rate": 0.00015336888128583,
      "loss": 10.277,
      "step": 488
    },
    {
      "epoch": 0.05359014226319719,
      "grad_norm": 0.05669292062520981,
      "learning_rate": 0.000153190455070635,
      "loss": 10.2875,
      "step": 489
    },
    {
      "epoch": 0.053699733556168965,
      "grad_norm": 0.0752144455909729,
      "learning_rate": 0.00015301179239376938,
      "loss": 10.2817,
      "step": 490
    },
    {
      "epoch": 0.053809324849140736,
      "grad_norm": 0.04608868062496185,
      "learning_rate": 0.00015283289404948976,
      "loss": 10.2861,
      "step": 491
    },
    {
      "epoch": 0.05391891614211251,
      "grad_norm": 0.05547158047556877,
      "learning_rate": 0.0001526537608331006,
      "loss": 10.2823,
      "step": 492
    },
    {
      "epoch": 0.054028507435084284,
      "grad_norm": 0.05423187091946602,
      "learning_rate": 0.00015247439354095041,
      "loss": 10.2836,
      "step": 493
    },
    {
      "epoch": 0.054138098728056055,
      "grad_norm": 0.05455704778432846,
      "learning_rate": 0.00015229479297042823,
      "loss": 10.2831,
      "step": 494
    },
    {
      "epoch": 0.054247690021027827,
      "grad_norm": 0.05773872882127762,
      "learning_rate": 0.00015211495991996027,
      "loss": 10.2893,
      "step": 495
    },
    {
      "epoch": 0.054357281313999604,
      "grad_norm": 0.04852631688117981,
      "learning_rate": 0.0001519348951890062,
      "loss": 10.2817,
      "step": 496
    },
    {
      "epoch": 0.054466872606971375,
      "grad_norm": 0.06247256323695183,
      "learning_rate": 0.0001517545995780556,
      "loss": 10.2744,
      "step": 497
    },
    {
      "epoch": 0.054576463899943146,
      "grad_norm": 0.05275551229715347,
      "learning_rate": 0.00015157407388862452,
      "loss": 10.2752,
      "step": 498
    },
    {
      "epoch": 0.054686055192914924,
      "grad_norm": 0.07741624116897583,
      "learning_rate": 0.00015139331892325179,
      "loss": 10.2929,
      "step": 499
    },
    {
      "epoch": 0.054795646485886695,
      "grad_norm": 0.061968181282281876,
      "learning_rate": 0.0001512123354854955,
      "loss": 10.2894,
      "step": 500
    },
    {
      "epoch": 0.054905237778858466,
      "grad_norm": 0.051437873393297195,
      "learning_rate": 0.0001510311243799295,
      "loss": 10.2853,
      "step": 501
    },
    {
      "epoch": 0.055014829071830244,
      "grad_norm": 0.04314221814274788,
      "learning_rate": 0.00015084968641213958,
      "loss": 10.2878,
      "step": 502
    },
    {
      "epoch": 0.055124420364802015,
      "grad_norm": 0.04410262405872345,
      "learning_rate": 0.00015066802238872023,
      "loss": 10.2829,
      "step": 503
    },
    {
      "epoch": 0.05523401165777379,
      "grad_norm": 0.046905409544706345,
      "learning_rate": 0.0001504861331172709,
      "loss": 10.2864,
      "step": 504
    },
    {
      "epoch": 0.055343602950745564,
      "grad_norm": 0.04072285816073418,
      "learning_rate": 0.0001503040194063922,
      "loss": 10.2855,
      "step": 505
    },
    {
      "epoch": 0.055453194243717335,
      "grad_norm": 0.059338562190532684,
      "learning_rate": 0.00015012168206568268,
      "loss": 10.287,
      "step": 506
    },
    {
      "epoch": 0.05556278553668911,
      "grad_norm": 0.0460449643433094,
      "learning_rate": 0.00014993912190573505,
      "loss": 10.2848,
      "step": 507
    },
    {
      "epoch": 0.055672376829660884,
      "grad_norm": 0.04710105061531067,
      "learning_rate": 0.00014975633973813242,
      "loss": 10.2868,
      "step": 508
    },
    {
      "epoch": 0.055781968122632655,
      "grad_norm": 0.05931161344051361,
      "learning_rate": 0.00014957333637544503,
      "loss": 10.2868,
      "step": 509
    },
    {
      "epoch": 0.05589155941560443,
      "grad_norm": 0.058749500662088394,
      "learning_rate": 0.00014939011263122634,
      "loss": 10.2818,
      "step": 510
    },
    {
      "epoch": 0.056001150708576204,
      "grad_norm": 0.054822489619255066,
      "learning_rate": 0.0001492066693200096,
      "loss": 10.2886,
      "step": 511
    },
    {
      "epoch": 0.056110742001547975,
      "grad_norm": 0.06286199390888214,
      "learning_rate": 0.00014902300725730413,
      "loss": 10.2848,
      "step": 512
    },
    {
      "epoch": 0.05622033329451975,
      "grad_norm": 0.0601000152528286,
      "learning_rate": 0.00014883912725959167,
      "loss": 10.2843,
      "step": 513
    },
    {
      "epoch": 0.056329924587491524,
      "grad_norm": 0.06388809531927109,
      "learning_rate": 0.00014865503014432292,
      "loss": 10.2937,
      "step": 514
    },
    {
      "epoch": 0.056439515880463295,
      "grad_norm": 0.06279318034648895,
      "learning_rate": 0.00014847071672991367,
      "loss": 10.2833,
      "step": 515
    },
    {
      "epoch": 0.05654910717343507,
      "grad_norm": 0.04898042604327202,
      "learning_rate": 0.0001482861878357414,
      "loss": 10.2848,
      "step": 516
    },
    {
      "epoch": 0.056658698466406844,
      "grad_norm": 0.055138904601335526,
      "learning_rate": 0.00014810144428214144,
      "loss": 10.2812,
      "step": 517
    },
    {
      "epoch": 0.056768289759378615,
      "grad_norm": 0.04735207185149193,
      "learning_rate": 0.0001479164868904034,
      "loss": 10.2832,
      "step": 518
    },
    {
      "epoch": 0.05687788105235039,
      "grad_norm": 0.058676231652498245,
      "learning_rate": 0.00014773131648276758,
      "loss": 10.2864,
      "step": 519
    },
    {
      "epoch": 0.056987472345322164,
      "grad_norm": 0.045081447809934616,
      "learning_rate": 0.00014754593388242117,
      "loss": 10.2743,
      "step": 520
    },
    {
      "epoch": 0.057097063638293935,
      "grad_norm": 0.0500495471060276,
      "learning_rate": 0.0001473603399134948,
      "loss": 10.2793,
      "step": 521
    },
    {
      "epoch": 0.05720665493126571,
      "grad_norm": 0.043647486716508865,
      "learning_rate": 0.0001471745354010586,
      "loss": 10.2891,
      "step": 522
    },
    {
      "epoch": 0.057316246224237484,
      "grad_norm": 0.045387063175439835,
      "learning_rate": 0.00014698852117111884,
      "loss": 10.2762,
      "step": 523
    },
    {
      "epoch": 0.057425837517209255,
      "grad_norm": 0.052651915699243546,
      "learning_rate": 0.000146802298050614,
      "loss": 10.2798,
      "step": 524
    },
    {
      "epoch": 0.05753542881018103,
      "grad_norm": 0.04874437302350998,
      "learning_rate": 0.0001466158668674112,
      "loss": 10.2894,
      "step": 525
    },
    {
      "epoch": 0.057645020103152804,
      "grad_norm": 0.04746442288160324,
      "learning_rate": 0.00014642922845030257,
      "loss": 10.2854,
      "step": 526
    },
    {
      "epoch": 0.057754611396124575,
      "grad_norm": 0.051118940114974976,
      "learning_rate": 0.0001462423836290015,
      "loss": 10.2786,
      "step": 527
    },
    {
      "epoch": 0.05786420268909635,
      "grad_norm": 0.0506085641682148,
      "learning_rate": 0.00014605533323413887,
      "loss": 10.2739,
      "step": 528
    },
    {
      "epoch": 0.057973793982068124,
      "grad_norm": 0.043574899435043335,
      "learning_rate": 0.00014586807809725962,
      "loss": 10.284,
      "step": 529
    },
    {
      "epoch": 0.058083385275039895,
      "grad_norm": 0.05374184995889664,
      "learning_rate": 0.00014568061905081875,
      "loss": 10.2802,
      "step": 530
    },
    {
      "epoch": 0.05819297656801167,
      "grad_norm": 0.06223062053322792,
      "learning_rate": 0.00014549295692817778,
      "loss": 10.2843,
      "step": 531
    },
    {
      "epoch": 0.058302567860983444,
      "grad_norm": 0.04860878735780716,
      "learning_rate": 0.00014530509256360102,
      "loss": 10.2745,
      "step": 532
    },
    {
      "epoch": 0.058412159153955215,
      "grad_norm": 0.04590313136577606,
      "learning_rate": 0.00014511702679225193,
      "loss": 10.2879,
      "step": 533
    },
    {
      "epoch": 0.05852175044692699,
      "grad_norm": 0.057339657098054886,
      "learning_rate": 0.0001449287604501893,
      "loss": 10.2808,
      "step": 534
    },
    {
      "epoch": 0.058631341739898764,
      "grad_norm": 0.048368241637945175,
      "learning_rate": 0.00014474029437436348,
      "loss": 10.2723,
      "step": 535
    },
    {
      "epoch": 0.05874093303287054,
      "grad_norm": 0.04476451873779297,
      "learning_rate": 0.00014455162940261285,
      "loss": 10.2723,
      "step": 536
    },
    {
      "epoch": 0.05885052432584231,
      "grad_norm": 0.07174571603536606,
      "learning_rate": 0.0001443627663736599,
      "loss": 10.2855,
      "step": 537
    },
    {
      "epoch": 0.058960115618814084,
      "grad_norm": 0.061415642499923706,
      "learning_rate": 0.00014417370612710778,
      "loss": 10.2866,
      "step": 538
    },
    {
      "epoch": 0.05906970691178586,
      "grad_norm": 0.052190665155649185,
      "learning_rate": 0.00014398444950343623,
      "loss": 10.2765,
      "step": 539
    },
    {
      "epoch": 0.05917929820475763,
      "grad_norm": 0.04251325502991676,
      "learning_rate": 0.00014379499734399798,
      "loss": 10.2799,
      "step": 540
    },
    {
      "epoch": 0.059288889497729404,
      "grad_norm": 0.05979350209236145,
      "learning_rate": 0.0001436053504910151,
      "loss": 10.2848,
      "step": 541
    },
    {
      "epoch": 0.05939848079070118,
      "grad_norm": 0.04662727192044258,
      "learning_rate": 0.0001434155097875752,
      "loss": 10.2807,
      "step": 542
    },
    {
      "epoch": 0.05950807208367295,
      "grad_norm": 0.06249745190143585,
      "learning_rate": 0.00014322547607762762,
      "loss": 10.278,
      "step": 543
    },
    {
      "epoch": 0.059617663376644724,
      "grad_norm": 0.061137039214372635,
      "learning_rate": 0.0001430352502059797,
      "loss": 10.288,
      "step": 544
    },
    {
      "epoch": 0.0597272546696165,
      "grad_norm": 0.06321033090353012,
      "learning_rate": 0.0001428448330182931,
      "loss": 10.2743,
      "step": 545
    },
    {
      "epoch": 0.05983684596258827,
      "grad_norm": 0.0559585765004158,
      "learning_rate": 0.00014265422536107993,
      "loss": 10.288,
      "step": 546
    },
    {
      "epoch": 0.059946437255560044,
      "grad_norm": 0.06237563490867615,
      "learning_rate": 0.00014246342808169914,
      "loss": 10.28,
      "step": 547
    },
    {
      "epoch": 0.06005602854853182,
      "grad_norm": 0.05317535251379013,
      "learning_rate": 0.00014227244202835257,
      "loss": 10.2782,
      "step": 548
    },
    {
      "epoch": 0.06016561984150359,
      "grad_norm": 0.059921957552433014,
      "learning_rate": 0.0001420812680500813,
      "loss": 10.2832,
      "step": 549
    },
    {
      "epoch": 0.060275211134475364,
      "grad_norm": 0.07936064153909683,
      "learning_rate": 0.00014188990699676184,
      "loss": 10.2879,
      "step": 550
    },
    {
      "epoch": 0.06038480242744714,
      "grad_norm": 0.04322652891278267,
      "learning_rate": 0.00014169835971910238,
      "loss": 10.2897,
      "step": 551
    },
    {
      "epoch": 0.06049439372041891,
      "grad_norm": 0.04569587856531143,
      "learning_rate": 0.0001415066270686389,
      "loss": 10.2893,
      "step": 552
    },
    {
      "epoch": 0.060603985013390684,
      "grad_norm": 0.04834531620144844,
      "learning_rate": 0.00014131470989773158,
      "loss": 10.2879,
      "step": 553
    },
    {
      "epoch": 0.06071357630636246,
      "grad_norm": 0.05462159961462021,
      "learning_rate": 0.0001411226090595608,
      "loss": 10.2821,
      "step": 554
    },
    {
      "epoch": 0.06082316759933423,
      "grad_norm": 0.05730734020471573,
      "learning_rate": 0.00014093032540812348,
      "loss": 10.2824,
      "step": 555
    },
    {
      "epoch": 0.060932758892306,
      "grad_norm": 0.05379139259457588,
      "learning_rate": 0.0001407378597982293,
      "loss": 10.2901,
      "step": 556
    },
    {
      "epoch": 0.06104235018527778,
      "grad_norm": 0.045758653432130814,
      "learning_rate": 0.00014054521308549673,
      "loss": 10.2924,
      "step": 557
    },
    {
      "epoch": 0.06115194147824955,
      "grad_norm": 0.05778197944164276,
      "learning_rate": 0.0001403523861263495,
      "loss": 10.2879,
      "step": 558
    },
    {
      "epoch": 0.06126153277122132,
      "grad_norm": 0.04310181364417076,
      "learning_rate": 0.00014015937977801256,
      "loss": 10.2868,
      "step": 559
    },
    {
      "epoch": 0.0613711240641931,
      "grad_norm": 0.055026717483997345,
      "learning_rate": 0.00013996619489850822,
      "loss": 10.287,
      "step": 560
    },
    {
      "epoch": 0.06148071535716487,
      "grad_norm": 0.05167794227600098,
      "learning_rate": 0.00013977283234665273,
      "loss": 10.2873,
      "step": 561
    },
    {
      "epoch": 0.06159030665013664,
      "grad_norm": 0.05238218232989311,
      "learning_rate": 0.00013957929298205195,
      "loss": 10.2843,
      "step": 562
    },
    {
      "epoch": 0.06169989794310842,
      "grad_norm": 0.042274102568626404,
      "learning_rate": 0.00013938557766509792,
      "loss": 10.282,
      "step": 563
    },
    {
      "epoch": 0.06180948923608019,
      "grad_norm": 0.0708811804652214,
      "learning_rate": 0.0001391916872569648,
      "loss": 10.28,
      "step": 564
    },
    {
      "epoch": 0.06191908052905197,
      "grad_norm": 0.08044921606779099,
      "learning_rate": 0.00013899762261960518,
      "loss": 10.2883,
      "step": 565
    },
    {
      "epoch": 0.06202867182202374,
      "grad_norm": 0.07139541953802109,
      "learning_rate": 0.0001388033846157462,
      "loss": 10.2861,
      "step": 566
    },
    {
      "epoch": 0.06213826311499551,
      "grad_norm": 0.041817497462034225,
      "learning_rate": 0.0001386089741088857,
      "loss": 10.2814,
      "step": 567
    },
    {
      "epoch": 0.06224785440796729,
      "grad_norm": 0.05387767031788826,
      "learning_rate": 0.00013841439196328836,
      "loss": 10.2771,
      "step": 568
    },
    {
      "epoch": 0.06235744570093906,
      "grad_norm": 0.04203671216964722,
      "learning_rate": 0.00013821963904398193,
      "loss": 10.2741,
      "step": 569
    },
    {
      "epoch": 0.06246703699391083,
      "grad_norm": 0.0553942508995533,
      "learning_rate": 0.00013802471621675338,
      "loss": 10.2753,
      "step": 570
    },
    {
      "epoch": 0.06257662828688261,
      "grad_norm": 0.0392155647277832,
      "learning_rate": 0.00013782962434814492,
      "loss": 10.28,
      "step": 571
    },
    {
      "epoch": 0.06268621957985437,
      "grad_norm": 0.051757361739873886,
      "learning_rate": 0.00013763436430545034,
      "loss": 10.2905,
      "step": 572
    },
    {
      "epoch": 0.06279581087282615,
      "grad_norm": 0.04969177767634392,
      "learning_rate": 0.00013743893695671096,
      "loss": 10.2871,
      "step": 573
    },
    {
      "epoch": 0.06290540216579793,
      "grad_norm": 0.046375542879104614,
      "learning_rate": 0.00013724334317071198,
      "loss": 10.2807,
      "step": 574
    },
    {
      "epoch": 0.0630149934587697,
      "grad_norm": 0.049421556293964386,
      "learning_rate": 0.00013704758381697844,
      "loss": 10.2838,
      "step": 575
    },
    {
      "epoch": 0.06312458475174147,
      "grad_norm": 0.057203248143196106,
      "learning_rate": 0.00013685165976577146,
      "loss": 10.2803,
      "step": 576
    },
    {
      "epoch": 0.06323417604471325,
      "grad_norm": 0.05681205913424492,
      "learning_rate": 0.0001366555718880843,
      "loss": 10.2872,
      "step": 577
    },
    {
      "epoch": 0.06334376733768501,
      "grad_norm": 0.05492445454001427,
      "learning_rate": 0.00013645932105563844,
      "loss": 10.2782,
      "step": 578
    },
    {
      "epoch": 0.06345335863065679,
      "grad_norm": 0.044668544083833694,
      "learning_rate": 0.00013626290814088005,
      "loss": 10.2869,
      "step": 579
    },
    {
      "epoch": 0.06356294992362857,
      "grad_norm": 0.04726185277104378,
      "learning_rate": 0.00013606633401697557,
      "loss": 10.2853,
      "step": 580
    },
    {
      "epoch": 0.06367254121660035,
      "grad_norm": 0.03690396994352341,
      "learning_rate": 0.00013586959955780824,
      "loss": 10.2788,
      "step": 581
    },
    {
      "epoch": 0.06378213250957211,
      "grad_norm": 0.04528044909238815,
      "learning_rate": 0.00013567270563797398,
      "loss": 10.2788,
      "step": 582
    },
    {
      "epoch": 0.06389172380254389,
      "grad_norm": 0.049076732248067856,
      "learning_rate": 0.00013547565313277776,
      "loss": 10.2831,
      "step": 583
    },
    {
      "epoch": 0.06400131509551567,
      "grad_norm": 0.0536038875579834,
      "learning_rate": 0.00013527844291822948,
      "loss": 10.2775,
      "step": 584
    },
    {
      "epoch": 0.06411090638848743,
      "grad_norm": 0.05230176821351051,
      "learning_rate": 0.0001350810758710401,
      "loss": 10.2824,
      "step": 585
    },
    {
      "epoch": 0.06422049768145921,
      "grad_norm": 0.055377472192049026,
      "learning_rate": 0.00013488355286861783,
      "loss": 10.2831,
      "step": 586
    },
    {
      "epoch": 0.06433008897443099,
      "grad_norm": 0.05220828205347061,
      "learning_rate": 0.0001346858747890642,
      "loss": 10.2802,
      "step": 587
    },
    {
      "epoch": 0.06443968026740275,
      "grad_norm": 0.06072499230504036,
      "learning_rate": 0.00013448804251117003,
      "loss": 10.2748,
      "step": 588
    },
    {
      "epoch": 0.06454927156037453,
      "grad_norm": 0.05654091387987137,
      "learning_rate": 0.0001342900569144119,
      "loss": 10.282,
      "step": 589
    },
    {
      "epoch": 0.06465886285334631,
      "grad_norm": 0.06162402406334877,
      "learning_rate": 0.0001340919188789477,
      "loss": 10.2649,
      "step": 590
    },
    {
      "epoch": 0.06476845414631807,
      "grad_norm": 0.050924353301525116,
      "learning_rate": 0.00013389362928561317,
      "loss": 10.2718,
      "step": 591
    },
    {
      "epoch": 0.06487804543928985,
      "grad_norm": 0.05021136999130249,
      "learning_rate": 0.00013369518901591772,
      "loss": 10.2846,
      "step": 592
    },
    {
      "epoch": 0.06498763673226163,
      "grad_norm": 0.053878482431173325,
      "learning_rate": 0.00013349659895204067,
      "loss": 10.2791,
      "step": 593
    },
    {
      "epoch": 0.06509722802523339,
      "grad_norm": 0.05243099108338356,
      "learning_rate": 0.0001332978599768272,
      "loss": 10.2761,
      "step": 594
    },
    {
      "epoch": 0.06520681931820517,
      "grad_norm": 0.05130923539400101,
      "learning_rate": 0.00013309897297378455,
      "loss": 10.2756,
      "step": 595
    },
    {
      "epoch": 0.06531641061117695,
      "grad_norm": 0.05565524473786354,
      "learning_rate": 0.00013289993882707797,
      "loss": 10.2727,
      "step": 596
    },
    {
      "epoch": 0.06542600190414871,
      "grad_norm": 0.04936867207288742,
      "learning_rate": 0.00013270075842152678,
      "loss": 10.2789,
      "step": 597
    },
    {
      "epoch": 0.06553559319712049,
      "grad_norm": 0.06208011880517006,
      "learning_rate": 0.00013250143264260074,
      "loss": 10.2769,
      "step": 598
    },
    {
      "epoch": 0.06564518449009227,
      "grad_norm": 0.06434591859579086,
      "learning_rate": 0.0001323019623764156,
      "loss": 10.2797,
      "step": 599
    },
    {
      "epoch": 0.06575477578306403,
      "grad_norm": 0.07852920889854431,
      "learning_rate": 0.00013210234850972964,
      "loss": 10.271,
      "step": 600
    },
    {
      "epoch": 0.06586436707603581,
      "grad_norm": 0.038199882954359055,
      "learning_rate": 0.0001319025919299394,
      "loss": 10.2877,
      "step": 601
    },
    {
      "epoch": 0.06597395836900759,
      "grad_norm": 0.041325367987155914,
      "learning_rate": 0.00013170269352507597,
      "loss": 10.288,
      "step": 602
    },
    {
      "epoch": 0.06608354966197935,
      "grad_norm": 0.0338250994682312,
      "learning_rate": 0.0001315026541838008,
      "loss": 10.2795,
      "step": 603
    },
    {
      "epoch": 0.06619314095495113,
      "grad_norm": 0.04901602864265442,
      "learning_rate": 0.00013130247479540202,
      "loss": 10.2781,
      "step": 604
    },
    {
      "epoch": 0.06630273224792291,
      "grad_norm": 0.05218203365802765,
      "learning_rate": 0.00013110215624979025,
      "loss": 10.2821,
      "step": 605
    },
    {
      "epoch": 0.06641232354089467,
      "grad_norm": 0.04709041118621826,
      "learning_rate": 0.00013090169943749476,
      "loss": 10.2855,
      "step": 606
    },
    {
      "epoch": 0.06652191483386645,
      "grad_norm": 0.05000196397304535,
      "learning_rate": 0.00013070110524965954,
      "loss": 10.2797,
      "step": 607
    },
    {
      "epoch": 0.06663150612683823,
      "grad_norm": 0.04985985532402992,
      "learning_rate": 0.00013050037457803924,
      "loss": 10.2709,
      "step": 608
    },
    {
      "epoch": 0.06674109741980999,
      "grad_norm": 0.044694121927022934,
      "learning_rate": 0.0001302995083149953,
      "loss": 10.2791,
      "step": 609
    },
    {
      "epoch": 0.06685068871278177,
      "grad_norm": 0.054903000593185425,
      "learning_rate": 0.0001300985073534919,
      "loss": 10.2838,
      "step": 610
    },
    {
      "epoch": 0.06696028000575355,
      "grad_norm": 0.05307765677571297,
      "learning_rate": 0.00012989737258709203,
      "loss": 10.2859,
      "step": 611
    },
    {
      "epoch": 0.06706987129872531,
      "grad_norm": 0.055976592004299164,
      "learning_rate": 0.00012969610490995358,
      "loss": 10.2905,
      "step": 612
    },
    {
      "epoch": 0.06717946259169709,
      "grad_norm": 0.055588845163583755,
      "learning_rate": 0.00012949470521682528,
      "loss": 10.279,
      "step": 613
    },
    {
      "epoch": 0.06728905388466887,
      "grad_norm": 0.05488145351409912,
      "learning_rate": 0.0001292931744030427,
      "loss": 10.2916,
      "step": 614
    },
    {
      "epoch": 0.06739864517764063,
      "grad_norm": 0.04581531509757042,
      "learning_rate": 0.0001290915133645243,
      "loss": 10.2805,
      "step": 615
    },
    {
      "epoch": 0.06750823647061241,
      "grad_norm": 0.05750222131609917,
      "learning_rate": 0.00012888972299776754,
      "loss": 10.2784,
      "step": 616
    },
    {
      "epoch": 0.06761782776358419,
      "grad_norm": 0.045469194650650024,
      "learning_rate": 0.00012868780419984482,
      "loss": 10.2653,
      "step": 617
    },
    {
      "epoch": 0.06772741905655595,
      "grad_norm": 0.046662114560604095,
      "learning_rate": 0.00012848575786839943,
      "loss": 10.2794,
      "step": 618
    },
    {
      "epoch": 0.06783701034952773,
      "grad_norm": 0.046689942479133606,
      "learning_rate": 0.0001282835849016416,
      "loss": 10.277,
      "step": 619
    },
    {
      "epoch": 0.06794660164249951,
      "grad_norm": 0.04662325605750084,
      "learning_rate": 0.00012808128619834461,
      "loss": 10.283,
      "step": 620
    },
    {
      "epoch": 0.06805619293547127,
      "grad_norm": 0.04100511968135834,
      "learning_rate": 0.0001278788626578407,
      "loss": 10.2812,
      "step": 621
    },
    {
      "epoch": 0.06816578422844305,
      "grad_norm": 0.04404404014348984,
      "learning_rate": 0.00012767631518001698,
      "loss": 10.2825,
      "step": 622
    },
    {
      "epoch": 0.06827537552141483,
      "grad_norm": 0.046444572508335114,
      "learning_rate": 0.00012747364466531163,
      "loss": 10.287,
      "step": 623
    },
    {
      "epoch": 0.06838496681438659,
      "grad_norm": 0.04805488511919975,
      "learning_rate": 0.00012727085201470973,
      "loss": 10.2821,
      "step": 624
    },
    {
      "epoch": 0.06849455810735837,
      "grad_norm": 0.04444395378232002,
      "learning_rate": 0.00012706793812973941,
      "loss": 10.2821,
      "step": 625
    },
    {
      "epoch": 0.06860414940033015,
      "grad_norm": 0.06284628808498383,
      "learning_rate": 0.0001268649039124677,
      "loss": 10.2836,
      "step": 626
    },
    {
      "epoch": 0.06871374069330191,
      "grad_norm": 0.04047633707523346,
      "learning_rate": 0.00012666175026549662,
      "loss": 10.2813,
      "step": 627
    },
    {
      "epoch": 0.06882333198627369,
      "grad_norm": 0.04667359218001366,
      "learning_rate": 0.000126458478091959,
      "loss": 10.2814,
      "step": 628
    },
    {
      "epoch": 0.06893292327924547,
      "grad_norm": 0.04300735518336296,
      "learning_rate": 0.00012625508829551473,
      "loss": 10.2853,
      "step": 629
    },
    {
      "epoch": 0.06904251457221723,
      "grad_norm": 0.058109112083911896,
      "learning_rate": 0.00012605158178034654,
      "loss": 10.2829,
      "step": 630
    },
    {
      "epoch": 0.06915210586518901,
      "grad_norm": 0.052021075040102005,
      "learning_rate": 0.00012584795945115603,
      "loss": 10.279,
      "step": 631
    },
    {
      "epoch": 0.06926169715816079,
      "grad_norm": 0.049847789108753204,
      "learning_rate": 0.0001256442222131597,
      "loss": 10.2693,
      "step": 632
    },
    {
      "epoch": 0.06937128845113255,
      "grad_norm": 0.05449390783905983,
      "learning_rate": 0.0001254403709720848,
      "loss": 10.2777,
      "step": 633
    },
    {
      "epoch": 0.06948087974410433,
      "grad_norm": 0.048660844564437866,
      "learning_rate": 0.0001252364066341655,
      "loss": 10.2833,
      "step": 634
    },
    {
      "epoch": 0.06959047103707611,
      "grad_norm": 0.04761822521686554,
      "learning_rate": 0.00012503233010613865,
      "loss": 10.269,
      "step": 635
    },
    {
      "epoch": 0.06970006233004787,
      "grad_norm": 0.05307494476437569,
      "learning_rate": 0.00012482814229523997,
      "loss": 10.2812,
      "step": 636
    },
    {
      "epoch": 0.06980965362301965,
      "grad_norm": 0.053977563977241516,
      "learning_rate": 0.00012462384410919975,
      "loss": 10.2761,
      "step": 637
    },
    {
      "epoch": 0.06991924491599143,
      "grad_norm": 0.07422066479921341,
      "learning_rate": 0.00012441943645623903,
      "loss": 10.2846,
      "step": 638
    },
    {
      "epoch": 0.07002883620896319,
      "grad_norm": 0.05061066895723343,
      "learning_rate": 0.00012421492024506555,
      "loss": 10.2774,
      "step": 639
    },
    {
      "epoch": 0.07013842750193497,
      "grad_norm": 0.048951782286167145,
      "learning_rate": 0.00012401029638486953,
      "loss": 10.2717,
      "step": 640
    },
    {
      "epoch": 0.07024801879490675,
      "grad_norm": 0.057149987667798996,
      "learning_rate": 0.0001238055657853198,
      "loss": 10.2784,
      "step": 641
    },
    {
      "epoch": 0.07035761008787852,
      "grad_norm": 0.042767126113176346,
      "learning_rate": 0.00012360072935655982,
      "loss": 10.2767,
      "step": 642
    },
    {
      "epoch": 0.07046720138085029,
      "grad_norm": 0.06343583017587662,
      "learning_rate": 0.00012339578800920332,
      "loss": 10.2717,
      "step": 643
    },
    {
      "epoch": 0.07057679267382207,
      "grad_norm": 0.056700509041547775,
      "learning_rate": 0.00012319074265433063,
      "loss": 10.2752,
      "step": 644
    },
    {
      "epoch": 0.07068638396679384,
      "grad_norm": 0.044347770512104034,
      "learning_rate": 0.00012298559420348437,
      "loss": 10.2783,
      "step": 645
    },
    {
      "epoch": 0.07079597525976561,
      "grad_norm": 0.05478415638208389,
      "learning_rate": 0.00012278034356866545,
      "loss": 10.2743,
      "step": 646
    },
    {
      "epoch": 0.07090556655273739,
      "grad_norm": 0.05334023758769035,
      "learning_rate": 0.00012257499166232907,
      "loss": 10.2763,
      "step": 647
    },
    {
      "epoch": 0.07101515784570916,
      "grad_norm": 0.062259215861558914,
      "learning_rate": 0.0001223695393973807,
      "loss": 10.2857,
      "step": 648
    },
    {
      "epoch": 0.07112474913868093,
      "grad_norm": 0.0609646737575531,
      "learning_rate": 0.0001221639876871719,
      "loss": 10.2719,
      "step": 649
    },
    {
      "epoch": 0.0712343404316527,
      "grad_norm": 0.0877390056848526,
      "learning_rate": 0.0001219583374454963,
      "loss": 10.2706,
      "step": 650
    },
    {
      "epoch": 0.07134393172462448,
      "grad_norm": 0.04235522076487541,
      "learning_rate": 0.00012175258958658564,
      "loss": 10.2921,
      "step": 651
    },
    {
      "epoch": 0.07145352301759625,
      "grad_norm": 0.04522537812590599,
      "learning_rate": 0.00012154674502510555,
      "loss": 10.2838,
      "step": 652
    },
    {
      "epoch": 0.07156311431056803,
      "grad_norm": 0.046782348304986954,
      "learning_rate": 0.00012134080467615159,
      "loss": 10.2849,
      "step": 653
    },
    {
      "epoch": 0.0716727056035398,
      "grad_norm": 0.05120275169610977,
      "learning_rate": 0.00012113476945524513,
      "loss": 10.2909,
      "step": 654
    },
    {
      "epoch": 0.07178229689651157,
      "grad_norm": 0.04147389903664589,
      "learning_rate": 0.00012092864027832933,
      "loss": 10.2902,
      "step": 655
    },
    {
      "epoch": 0.07189188818948335,
      "grad_norm": 0.045331697911024094,
      "learning_rate": 0.000120722418061765,
      "loss": 10.2827,
      "step": 656
    },
    {
      "epoch": 0.07200147948245512,
      "grad_norm": 0.04676318168640137,
      "learning_rate": 0.0001205161037223266,
      "loss": 10.2847,
      "step": 657
    },
    {
      "epoch": 0.07211107077542689,
      "grad_norm": 0.04648913815617561,
      "learning_rate": 0.00012030969817719808,
      "loss": 10.2878,
      "step": 658
    },
    {
      "epoch": 0.07222066206839867,
      "grad_norm": 0.05666414648294449,
      "learning_rate": 0.00012010320234396894,
      "loss": 10.294,
      "step": 659
    },
    {
      "epoch": 0.07233025336137044,
      "grad_norm": 0.050731558352708817,
      "learning_rate": 0.00011989661714062999,
      "loss": 10.2777,
      "step": 660
    },
    {
      "epoch": 0.07243984465434221,
      "grad_norm": 0.05709559842944145,
      "learning_rate": 0.0001196899434855693,
      "loss": 10.2726,
      "step": 661
    },
    {
      "epoch": 0.07254943594731399,
      "grad_norm": 0.054934509098529816,
      "learning_rate": 0.00011948318229756827,
      "loss": 10.2812,
      "step": 662
    },
    {
      "epoch": 0.07265902724028576,
      "grad_norm": 0.04980732873082161,
      "learning_rate": 0.00011927633449579735,
      "loss": 10.2779,
      "step": 663
    },
    {
      "epoch": 0.07276861853325753,
      "grad_norm": 0.060141194611787796,
      "learning_rate": 0.0001190694009998121,
      "loss": 10.2758,
      "step": 664
    },
    {
      "epoch": 0.0728782098262293,
      "grad_norm": 0.06462522596120834,
      "learning_rate": 0.00011886238272954897,
      "loss": 10.2723,
      "step": 665
    },
    {
      "epoch": 0.07298780111920108,
      "grad_norm": 0.05180748552083969,
      "learning_rate": 0.00011865528060532127,
      "loss": 10.2908,
      "step": 666
    },
    {
      "epoch": 0.07309739241217285,
      "grad_norm": 0.04206019639968872,
      "learning_rate": 0.0001184480955478152,
      "loss": 10.2747,
      "step": 667
    },
    {
      "epoch": 0.07320698370514463,
      "grad_norm": 0.0519523099064827,
      "learning_rate": 0.00011824082847808558,
      "loss": 10.2826,
      "step": 668
    },
    {
      "epoch": 0.0733165749981164,
      "grad_norm": 0.04755115881562233,
      "learning_rate": 0.00011803348031755179,
      "loss": 10.2858,
      "step": 669
    },
    {
      "epoch": 0.07342616629108817,
      "grad_norm": 0.03691481426358223,
      "learning_rate": 0.0001178260519879937,
      "loss": 10.2779,
      "step": 670
    },
    {
      "epoch": 0.07353575758405995,
      "grad_norm": 0.04195776954293251,
      "learning_rate": 0.00011761854441154767,
      "loss": 10.2834,
      "step": 671
    },
    {
      "epoch": 0.07364534887703172,
      "grad_norm": 0.05043170228600502,
      "learning_rate": 0.00011741095851070228,
      "loss": 10.2752,
      "step": 672
    },
    {
      "epoch": 0.07375494017000349,
      "grad_norm": 0.06182184815406799,
      "learning_rate": 0.00011720329520829429,
      "loss": 10.2816,
      "step": 673
    },
    {
      "epoch": 0.07386453146297527,
      "grad_norm": 0.046354733407497406,
      "learning_rate": 0.0001169955554275046,
      "loss": 10.2741,
      "step": 674
    },
    {
      "epoch": 0.07397412275594704,
      "grad_norm": 0.06240877881646156,
      "learning_rate": 0.0001167877400918541,
      "loss": 10.2688,
      "step": 675
    },
    {
      "epoch": 0.07408371404891881,
      "grad_norm": 0.05693672224879265,
      "learning_rate": 0.00011657985012519952,
      "loss": 10.275,
      "step": 676
    },
    {
      "epoch": 0.07419330534189059,
      "grad_norm": 0.05353149399161339,
      "learning_rate": 0.00011637188645172944,
      "loss": 10.2772,
      "step": 677
    },
    {
      "epoch": 0.07430289663486236,
      "grad_norm": 0.04571222886443138,
      "learning_rate": 0.00011616384999596006,
      "loss": 10.2843,
      "step": 678
    },
    {
      "epoch": 0.07441248792783413,
      "grad_norm": 0.05002811923623085,
      "learning_rate": 0.00011595574168273111,
      "loss": 10.2821,
      "step": 679
    },
    {
      "epoch": 0.0745220792208059,
      "grad_norm": 0.047933436930179596,
      "learning_rate": 0.0001157475624372018,
      "loss": 10.2875,
      "step": 680
    },
    {
      "epoch": 0.07463167051377768,
      "grad_norm": 0.0639643520116806,
      "learning_rate": 0.0001155393131848467,
      "loss": 10.2784,
      "step": 681
    },
    {
      "epoch": 0.07474126180674945,
      "grad_norm": 0.0475531704723835,
      "learning_rate": 0.00011533099485145155,
      "loss": 10.2765,
      "step": 682
    },
    {
      "epoch": 0.07485085309972123,
      "grad_norm": 0.04646487534046173,
      "learning_rate": 0.00011512260836310924,
      "loss": 10.2782,
      "step": 683
    },
    {
      "epoch": 0.074960444392693,
      "grad_norm": 0.06382658332586288,
      "learning_rate": 0.00011491415464621562,
      "loss": 10.2845,
      "step": 684
    },
    {
      "epoch": 0.07507003568566477,
      "grad_norm": 0.05752754211425781,
      "learning_rate": 0.00011470563462746541,
      "loss": 10.2788,
      "step": 685
    },
    {
      "epoch": 0.07517962697863655,
      "grad_norm": 0.060304757207632065,
      "learning_rate": 0.00011449704923384812,
      "loss": 10.2827,
      "step": 686
    },
    {
      "epoch": 0.07528921827160832,
      "grad_norm": 0.03880226984620094,
      "learning_rate": 0.00011428839939264382,
      "loss": 10.2779,
      "step": 687
    },
    {
      "epoch": 0.07539880956458009,
      "grad_norm": 0.04644261673092842,
      "learning_rate": 0.0001140796860314191,
      "loss": 10.2797,
      "step": 688
    },
    {
      "epoch": 0.07550840085755187,
      "grad_norm": 0.053935907781124115,
      "learning_rate": 0.00011387091007802297,
      "loss": 10.2844,
      "step": 689
    },
    {
      "epoch": 0.07561799215052364,
      "grad_norm": 0.07063944637775421,
      "learning_rate": 0.0001136620724605827,
      "loss": 10.2796,
      "step": 690
    },
    {
      "epoch": 0.07572758344349541,
      "grad_norm": 0.046229638159275055,
      "learning_rate": 0.00011345317410749964,
      "loss": 10.2698,
      "step": 691
    },
    {
      "epoch": 0.07583717473646719,
      "grad_norm": 0.056216951459646225,
      "learning_rate": 0.00011324421594744516,
      "loss": 10.2777,
      "step": 692
    },
    {
      "epoch": 0.07594676602943896,
      "grad_norm": 0.054789949208498,
      "learning_rate": 0.00011303519890935656,
      "loss": 10.2756,
      "step": 693
    },
    {
      "epoch": 0.07605635732241073,
      "grad_norm": 0.054642800241708755,
      "learning_rate": 0.00011282612392243286,
      "loss": 10.2775,
      "step": 694
    },
    {
      "epoch": 0.0761659486153825,
      "grad_norm": 0.058396000415086746,
      "learning_rate": 0.00011261699191613066,
      "loss": 10.2878,
      "step": 695
    },
    {
      "epoch": 0.07627553990835428,
      "grad_norm": 0.06086867302656174,
      "learning_rate": 0.00011240780382016005,
      "loss": 10.2751,
      "step": 696
    },
    {
      "epoch": 0.07638513120132605,
      "grad_norm": 0.06708884239196777,
      "learning_rate": 0.00011219856056448051,
      "loss": 10.2658,
      "step": 697
    },
    {
      "epoch": 0.07649472249429783,
      "grad_norm": 0.04813341796398163,
      "learning_rate": 0.00011198926307929664,
      "loss": 10.2773,
      "step": 698
    },
    {
      "epoch": 0.0766043137872696,
      "grad_norm": 0.050262704491615295,
      "learning_rate": 0.00011177991229505431,
      "loss": 10.2796,
      "step": 699
    },
    {
      "epoch": 0.07671390508024137,
      "grad_norm": 0.07048717141151428,
      "learning_rate": 0.00011157050914243614,
      "loss": 10.2899,
      "step": 700
    },
    {
      "epoch": 0.07682349637321315,
      "grad_norm": 0.03620591387152672,
      "learning_rate": 0.00011136105455235766,
      "loss": 10.2821,
      "step": 701
    },
    {
      "epoch": 0.07693308766618492,
      "grad_norm": 0.049698784947395325,
      "learning_rate": 0.00011115154945596305,
      "loss": 10.2889,
      "step": 702
    },
    {
      "epoch": 0.0770426789591567,
      "grad_norm": 0.0507047176361084,
      "learning_rate": 0.00011094199478462095,
      "loss": 10.2895,
      "step": 703
    },
    {
      "epoch": 0.07715227025212847,
      "grad_norm": 0.04329587146639824,
      "learning_rate": 0.00011073239146992054,
      "loss": 10.2972,
      "step": 704
    },
    {
      "epoch": 0.07726186154510024,
      "grad_norm": 0.04894951730966568,
      "learning_rate": 0.00011052274044366711,
      "loss": 10.2837,
      "step": 705
    },
    {
      "epoch": 0.07737145283807202,
      "grad_norm": 0.05422523245215416,
      "learning_rate": 0.00011031304263787812,
      "loss": 10.286,
      "step": 706
    },
    {
      "epoch": 0.07748104413104379,
      "grad_norm": 0.05212870612740517,
      "learning_rate": 0.00011010329898477891,
      "loss": 10.2789,
      "step": 707
    },
    {
      "epoch": 0.07759063542401556,
      "grad_norm": 0.051065925508737564,
      "learning_rate": 0.0001098935104167988,
      "loss": 10.2792,
      "step": 708
    },
    {
      "epoch": 0.07770022671698734,
      "grad_norm": 0.05591699853539467,
      "learning_rate": 0.00010968367786656663,
      "loss": 10.2868,
      "step": 709
    },
    {
      "epoch": 0.0778098180099591,
      "grad_norm": 0.05555687099695206,
      "learning_rate": 0.00010947380226690684,
      "loss": 10.2807,
      "step": 710
    },
    {
      "epoch": 0.07791940930293088,
      "grad_norm": 0.04770127311348915,
      "learning_rate": 0.00010926388455083522,
      "loss": 10.2817,
      "step": 711
    },
    {
      "epoch": 0.07802900059590266,
      "grad_norm": 0.044684287160634995,
      "learning_rate": 0.00010905392565155477,
      "loss": 10.2763,
      "step": 712
    },
    {
      "epoch": 0.07813859188887443,
      "grad_norm": 0.04942325875163078,
      "learning_rate": 0.00010884392650245165,
      "loss": 10.284,
      "step": 713
    },
    {
      "epoch": 0.0782481831818462,
      "grad_norm": 0.046786803752183914,
      "learning_rate": 0.00010863388803709089,
      "loss": 10.2802,
      "step": 714
    },
    {
      "epoch": 0.07835777447481798,
      "grad_norm": 0.06246323138475418,
      "learning_rate": 0.00010842381118921232,
      "loss": 10.278,
      "step": 715
    },
    {
      "epoch": 0.07846736576778975,
      "grad_norm": 0.06509517133235931,
      "learning_rate": 0.00010821369689272638,
      "loss": 10.2915,
      "step": 716
    },
    {
      "epoch": 0.07857695706076152,
      "grad_norm": 0.04616408795118332,
      "learning_rate": 0.00010800354608171003,
      "loss": 10.2784,
      "step": 717
    },
    {
      "epoch": 0.0786865483537333,
      "grad_norm": 0.04160367697477341,
      "learning_rate": 0.00010779335969040252,
      "loss": 10.2805,
      "step": 718
    },
    {
      "epoch": 0.07879613964670507,
      "grad_norm": 0.04817226156592369,
      "learning_rate": 0.00010758313865320134,
      "loss": 10.2838,
      "step": 719
    },
    {
      "epoch": 0.07890573093967684,
      "grad_norm": 0.03924615681171417,
      "learning_rate": 0.00010737288390465792,
      "loss": 10.2856,
      "step": 720
    },
    {
      "epoch": 0.07901532223264862,
      "grad_norm": 0.04477131366729736,
      "learning_rate": 0.00010716259637947357,
      "loss": 10.2823,
      "step": 721
    },
    {
      "epoch": 0.07912491352562039,
      "grad_norm": 0.04401953145861626,
      "learning_rate": 0.00010695227701249537,
      "loss": 10.2812,
      "step": 722
    },
    {
      "epoch": 0.07923450481859216,
      "grad_norm": 0.04418390244245529,
      "learning_rate": 0.00010674192673871191,
      "loss": 10.2772,
      "step": 723
    },
    {
      "epoch": 0.07934409611156394,
      "grad_norm": 0.046923235058784485,
      "learning_rate": 0.00010653154649324917,
      "loss": 10.276,
      "step": 724
    },
    {
      "epoch": 0.0794536874045357,
      "grad_norm": 0.04393209517002106,
      "learning_rate": 0.00010632113721136636,
      "loss": 10.2806,
      "step": 725
    },
    {
      "epoch": 0.07956327869750748,
      "grad_norm": 0.04967564344406128,
      "learning_rate": 0.00010611069982845183,
      "loss": 10.2745,
      "step": 726
    },
    {
      "epoch": 0.07967286999047926,
      "grad_norm": 0.062477946281433105,
      "learning_rate": 0.00010590023528001884,
      "loss": 10.2741,
      "step": 727
    },
    {
      "epoch": 0.07978246128345103,
      "grad_norm": 0.056707169860601425,
      "learning_rate": 0.00010568974450170139,
      "loss": 10.2887,
      "step": 728
    },
    {
      "epoch": 0.0798920525764228,
      "grad_norm": 0.04442475736141205,
      "learning_rate": 0.00010547922842925008,
      "loss": 10.2718,
      "step": 729
    },
    {
      "epoch": 0.08000164386939458,
      "grad_norm": 0.04830038547515869,
      "learning_rate": 0.00010526868799852796,
      "loss": 10.2778,
      "step": 730
    },
    {
      "epoch": 0.08011123516236635,
      "grad_norm": 0.04899175092577934,
      "learning_rate": 0.0001050581241455064,
      "loss": 10.2739,
      "step": 731
    },
    {
      "epoch": 0.08022082645533812,
      "grad_norm": 0.049212872982025146,
      "learning_rate": 0.00010484753780626089,
      "loss": 10.272,
      "step": 732
    },
    {
      "epoch": 0.0803304177483099,
      "grad_norm": 0.05184762924909592,
      "learning_rate": 0.00010463692991696685,
      "loss": 10.2793,
      "step": 733
    },
    {
      "epoch": 0.08044000904128167,
      "grad_norm": 0.055443175137043,
      "learning_rate": 0.00010442630141389549,
      "loss": 10.2768,
      "step": 734
    },
    {
      "epoch": 0.08054960033425344,
      "grad_norm": 0.04665730148553848,
      "learning_rate": 0.00010421565323340971,
      "loss": 10.2867,
      "step": 735
    },
    {
      "epoch": 0.08065919162722522,
      "grad_norm": 0.06016243249177933,
      "learning_rate": 0.00010400498631195992,
      "loss": 10.2883,
      "step": 736
    },
    {
      "epoch": 0.08076878292019699,
      "grad_norm": 0.04775187373161316,
      "learning_rate": 0.00010379430158607975,
      "loss": 10.2721,
      "step": 737
    },
    {
      "epoch": 0.08087837421316876,
      "grad_norm": 0.06185443699359894,
      "learning_rate": 0.000103583599992382,
      "loss": 10.2735,
      "step": 738
    },
    {
      "epoch": 0.08098796550614054,
      "grad_norm": 0.05255201831459999,
      "learning_rate": 0.0001033728824675545,
      "loss": 10.2712,
      "step": 739
    },
    {
      "epoch": 0.0810975567991123,
      "grad_norm": 0.06066211685538292,
      "learning_rate": 0.0001031621499483559,
      "loss": 10.2687,
      "step": 740
    },
    {
      "epoch": 0.08120714809208408,
      "grad_norm": 0.05449224263429642,
      "learning_rate": 0.00010295140337161146,
      "loss": 10.2739,
      "step": 741
    },
    {
      "epoch": 0.08131673938505586,
      "grad_norm": 0.05965689942240715,
      "learning_rate": 0.00010274064367420897,
      "loss": 10.2772,
      "step": 742
    },
    {
      "epoch": 0.08142633067802763,
      "grad_norm": 0.05739325284957886,
      "learning_rate": 0.00010252987179309459,
      "loss": 10.2748,
      "step": 743
    },
    {
      "epoch": 0.0815359219709994,
      "grad_norm": 0.051967184990644455,
      "learning_rate": 0.00010231908866526851,
      "loss": 10.2784,
      "step": 744
    },
    {
      "epoch": 0.08164551326397118,
      "grad_norm": 0.060379285365343094,
      "learning_rate": 0.00010210829522778111,
      "loss": 10.2789,
      "step": 745
    },
    {
      "epoch": 0.08175510455694295,
      "grad_norm": 0.06187751144170761,
      "learning_rate": 0.00010189749241772844,
      "loss": 10.2753,
      "step": 746
    },
    {
      "epoch": 0.08186469584991472,
      "grad_norm": 0.06870713829994202,
      "learning_rate": 0.00010168668117224825,
      "loss": 10.269,
      "step": 747
    },
    {
      "epoch": 0.0819742871428865,
      "grad_norm": 0.05520250275731087,
      "learning_rate": 0.00010147586242851585,
      "loss": 10.2688,
      "step": 748
    },
    {
      "epoch": 0.08208387843585827,
      "grad_norm": 0.055792663246393204,
      "learning_rate": 0.00010126503712373982,
      "loss": 10.2755,
      "step": 749
    },
    {
      "epoch": 0.08219346972883004,
      "grad_norm": 0.0639624372124672,
      "learning_rate": 0.00010105420619515798,
      "loss": 10.268,
      "step": 750
    },
    {
      "epoch": 0.08219346972883004,
      "eval_loss": 10.27712631225586,
      "eval_runtime": 55.6,
      "eval_samples_per_second": 276.421,
      "eval_steps_per_second": 138.219,
      "step": 750
    },
    {
      "epoch": 0.08230306102180182,
      "grad_norm": 0.043025027960538864,
      "learning_rate": 0.00010084337058003303,
      "loss": 10.2856,
      "step": 751
    },
    {
      "epoch": 0.08241265231477359,
      "grad_norm": 0.04668436199426651,
      "learning_rate": 0.00010063253121564868,
      "loss": 10.2842,
      "step": 752
    },
    {
      "epoch": 0.08252224360774536,
      "grad_norm": 0.03862539306282997,
      "learning_rate": 0.00010042168903930514,
      "loss": 10.279,
      "step": 753
    },
    {
      "epoch": 0.08263183490071714,
      "grad_norm": 0.03826581686735153,
      "learning_rate": 0.00010021084498831522,
      "loss": 10.2935,
      "step": 754
    },
    {
      "epoch": 0.0827414261936889,
      "grad_norm": 0.047461580485105515,
      "learning_rate": 0.0001,
      "loss": 10.2828,
      "step": 755
    },
    {
      "epoch": 0.08285101748666068,
      "grad_norm": 0.05469823256134987,
      "learning_rate": 9.97891550116848e-05,
      "loss": 10.2816,
      "step": 756
    },
    {
      "epoch": 0.08296060877963246,
      "grad_norm": 0.05587458983063698,
      "learning_rate": 9.957831096069488e-05,
      "loss": 10.2795,
      "step": 757
    },
    {
      "epoch": 0.08307020007260423,
      "grad_norm": 0.05442638322710991,
      "learning_rate": 9.936746878435136e-05,
      "loss": 10.2823,
      "step": 758
    },
    {
      "epoch": 0.083179791365576,
      "grad_norm": 0.053162962198257446,
      "learning_rate": 9.915662941996699e-05,
      "loss": 10.2815,
      "step": 759
    },
    {
      "epoch": 0.08328938265854778,
      "grad_norm": 0.04670391604304314,
      "learning_rate": 9.894579380484204e-05,
      "loss": 10.2802,
      "step": 760
    },
    {
      "epoch": 0.08339897395151954,
      "grad_norm": 0.05810031667351723,
      "learning_rate": 9.873496287626019e-05,
      "loss": 10.2838,
      "step": 761
    },
    {
      "epoch": 0.08350856524449132,
      "grad_norm": 0.05600173771381378,
      "learning_rate": 9.852413757148417e-05,
      "loss": 10.2786,
      "step": 762
    },
    {
      "epoch": 0.0836181565374631,
      "grad_norm": 0.054038550704717636,
      "learning_rate": 9.831331882775178e-05,
      "loss": 10.2791,
      "step": 763
    },
    {
      "epoch": 0.08372774783043488,
      "grad_norm": 0.06092235445976257,
      "learning_rate": 9.81025075822716e-05,
      "loss": 10.2802,
      "step": 764
    },
    {
      "epoch": 0.08383733912340664,
      "grad_norm": 0.05696984380483627,
      "learning_rate": 9.789170477221891e-05,
      "loss": 10.2845,
      "step": 765
    },
    {
      "epoch": 0.08394693041637842,
      "grad_norm": 0.05985541641712189,
      "learning_rate": 9.76809113347315e-05,
      "loss": 10.2797,
      "step": 766
    },
    {
      "epoch": 0.0840565217093502,
      "grad_norm": 0.05183318629860878,
      "learning_rate": 9.747012820690543e-05,
      "loss": 10.28,
      "step": 767
    },
    {
      "epoch": 0.08416611300232196,
      "grad_norm": 0.038727641105651855,
      "learning_rate": 9.725935632579104e-05,
      "loss": 10.2773,
      "step": 768
    },
    {
      "epoch": 0.08427570429529374,
      "grad_norm": 0.05454067140817642,
      "learning_rate": 9.704859662838855e-05,
      "loss": 10.2765,
      "step": 769
    },
    {
      "epoch": 0.08438529558826552,
      "grad_norm": 0.04756341502070427,
      "learning_rate": 9.683785005164411e-05,
      "loss": 10.2785,
      "step": 770
    },
    {
      "epoch": 0.08449488688123728,
      "grad_norm": 0.04645154997706413,
      "learning_rate": 9.662711753244551e-05,
      "loss": 10.2692,
      "step": 771
    },
    {
      "epoch": 0.08460447817420906,
      "grad_norm": 0.04743330180644989,
      "learning_rate": 9.641640000761802e-05,
      "loss": 10.2782,
      "step": 772
    },
    {
      "epoch": 0.08471406946718084,
      "grad_norm": 0.053522802889347076,
      "learning_rate": 9.620569841392029e-05,
      "loss": 10.2775,
      "step": 773
    },
    {
      "epoch": 0.0848236607601526,
      "grad_norm": 0.0471879318356514,
      "learning_rate": 9.59950136880401e-05,
      "loss": 10.2818,
      "step": 774
    },
    {
      "epoch": 0.08493325205312438,
      "grad_norm": 0.048917997628450394,
      "learning_rate": 9.57843467665903e-05,
      "loss": 10.2801,
      "step": 775
    },
    {
      "epoch": 0.08504284334609616,
      "grad_norm": 0.05315422639250755,
      "learning_rate": 9.557369858610453e-05,
      "loss": 10.2776,
      "step": 776
    },
    {
      "epoch": 0.08515243463906792,
      "grad_norm": 0.04271043837070465,
      "learning_rate": 9.53630700830332e-05,
      "loss": 10.2837,
      "step": 777
    },
    {
      "epoch": 0.0852620259320397,
      "grad_norm": 0.05258758366107941,
      "learning_rate": 9.51524621937391e-05,
      "loss": 10.2765,
      "step": 778
    },
    {
      "epoch": 0.08537161722501148,
      "grad_norm": 0.043475668877363205,
      "learning_rate": 9.494187585449358e-05,
      "loss": 10.273,
      "step": 779
    },
    {
      "epoch": 0.08548120851798324,
      "grad_norm": 0.04373582452535629,
      "learning_rate": 9.473131200147205e-05,
      "loss": 10.2749,
      "step": 780
    },
    {
      "epoch": 0.08559079981095502,
      "grad_norm": 0.05771813541650772,
      "learning_rate": 9.452077157074994e-05,
      "loss": 10.2768,
      "step": 781
    },
    {
      "epoch": 0.0857003911039268,
      "grad_norm": 0.04305161163210869,
      "learning_rate": 9.431025549829862e-05,
      "loss": 10.2747,
      "step": 782
    },
    {
      "epoch": 0.08580998239689856,
      "grad_norm": 0.05582757666707039,
      "learning_rate": 9.409976471998118e-05,
      "loss": 10.2759,
      "step": 783
    },
    {
      "epoch": 0.08591957368987034,
      "grad_norm": 0.05292634665966034,
      "learning_rate": 9.388930017154819e-05,
      "loss": 10.2696,
      "step": 784
    },
    {
      "epoch": 0.08602916498284212,
      "grad_norm": 0.05195935443043709,
      "learning_rate": 9.367886278863366e-05,
      "loss": 10.2742,
      "step": 785
    },
    {
      "epoch": 0.08613875627581388,
      "grad_norm": 0.04931614547967911,
      "learning_rate": 9.346845350675088e-05,
      "loss": 10.2795,
      "step": 786
    },
    {
      "epoch": 0.08624834756878566,
      "grad_norm": 0.04614530876278877,
      "learning_rate": 9.325807326128814e-05,
      "loss": 10.2771,
      "step": 787
    },
    {
      "epoch": 0.08635793886175744,
      "grad_norm": 0.057682473212480545,
      "learning_rate": 9.304772298750463e-05,
      "loss": 10.276,
      "step": 788
    },
    {
      "epoch": 0.0864675301547292,
      "grad_norm": 0.05022144317626953,
      "learning_rate": 9.283740362052642e-05,
      "loss": 10.2761,
      "step": 789
    },
    {
      "epoch": 0.08657712144770098,
      "grad_norm": 0.04770553112030029,
      "learning_rate": 9.26271160953421e-05,
      "loss": 10.2621,
      "step": 790
    },
    {
      "epoch": 0.08668671274067276,
      "grad_norm": 0.05464399605989456,
      "learning_rate": 9.241686134679867e-05,
      "loss": 10.2715,
      "step": 791
    },
    {
      "epoch": 0.08679630403364452,
      "grad_norm": 0.04219220578670502,
      "learning_rate": 9.220664030959749e-05,
      "loss": 10.2752,
      "step": 792
    },
    {
      "epoch": 0.0869058953266163,
      "grad_norm": 0.04700864851474762,
      "learning_rate": 9.199645391828999e-05,
      "loss": 10.276,
      "step": 793
    },
    {
      "epoch": 0.08701548661958808,
      "grad_norm": 0.04892566055059433,
      "learning_rate": 9.178630310727365e-05,
      "loss": 10.273,
      "step": 794
    },
    {
      "epoch": 0.08712507791255984,
      "grad_norm": 0.054683417081832886,
      "learning_rate": 9.157618881078772e-05,
      "loss": 10.2811,
      "step": 795
    },
    {
      "epoch": 0.08723466920553162,
      "grad_norm": 0.06820336729288101,
      "learning_rate": 9.136611196290915e-05,
      "loss": 10.2726,
      "step": 796
    },
    {
      "epoch": 0.0873442604985034,
      "grad_norm": 0.061943039298057556,
      "learning_rate": 9.115607349754834e-05,
      "loss": 10.2708,
      "step": 797
    },
    {
      "epoch": 0.08745385179147516,
      "grad_norm": 0.043984200805425644,
      "learning_rate": 9.094607434844523e-05,
      "loss": 10.2719,
      "step": 798
    },
    {
      "epoch": 0.08756344308444694,
      "grad_norm": 0.06595799326896667,
      "learning_rate": 9.07361154491648e-05,
      "loss": 10.2773,
      "step": 799
    },
    {
      "epoch": 0.08767303437741872,
      "grad_norm": 0.053098298609256744,
      "learning_rate": 9.052619773309317e-05,
      "loss": 10.2821,
      "step": 800
    },
    {
      "epoch": 0.08778262567039048,
      "grad_norm": 0.043164510279893875,
      "learning_rate": 9.031632213343339e-05,
      "loss": 10.2817,
      "step": 801
    },
    {
      "epoch": 0.08789221696336226,
      "grad_norm": 0.045928992331027985,
      "learning_rate": 9.01064895832012e-05,
      "loss": 10.2827,
      "step": 802
    },
    {
      "epoch": 0.08800180825633404,
      "grad_norm": 0.053251467645168304,
      "learning_rate": 8.98967010152211e-05,
      "loss": 10.2833,
      "step": 803
    },
    {
      "epoch": 0.0881113995493058,
      "grad_norm": 0.04367321729660034,
      "learning_rate": 8.968695736212193e-05,
      "loss": 10.2795,
      "step": 804
    },
    {
      "epoch": 0.08822099084227758,
      "grad_norm": 0.04833493381738663,
      "learning_rate": 8.947725955633294e-05,
      "loss": 10.2759,
      "step": 805
    },
    {
      "epoch": 0.08833058213524936,
      "grad_norm": 0.0468059778213501,
      "learning_rate": 8.926760853007946e-05,
      "loss": 10.2806,
      "step": 806
    },
    {
      "epoch": 0.08844017342822112,
      "grad_norm": 0.04794561490416527,
      "learning_rate": 8.905800521537905e-05,
      "loss": 10.2729,
      "step": 807
    },
    {
      "epoch": 0.0885497647211929,
      "grad_norm": 0.04558403789997101,
      "learning_rate": 8.884845054403699e-05,
      "loss": 10.2855,
      "step": 808
    },
    {
      "epoch": 0.08865935601416468,
      "grad_norm": 0.04180232062935829,
      "learning_rate": 8.863894544764236e-05,
      "loss": 10.2779,
      "step": 809
    },
    {
      "epoch": 0.08876894730713644,
      "grad_norm": 0.04768197983503342,
      "learning_rate": 8.84294908575639e-05,
      "loss": 10.2772,
      "step": 810
    },
    {
      "epoch": 0.08887853860010822,
      "grad_norm": 0.04308689758181572,
      "learning_rate": 8.822008770494572e-05,
      "loss": 10.2805,
      "step": 811
    },
    {
      "epoch": 0.08898812989308,
      "grad_norm": 0.04822627082467079,
      "learning_rate": 8.801073692070337e-05,
      "loss": 10.273,
      "step": 812
    },
    {
      "epoch": 0.08909772118605176,
      "grad_norm": 0.057119715958833694,
      "learning_rate": 8.780143943551954e-05,
      "loss": 10.2861,
      "step": 813
    },
    {
      "epoch": 0.08920731247902354,
      "grad_norm": 0.056585513055324554,
      "learning_rate": 8.759219617983999e-05,
      "loss": 10.2934,
      "step": 814
    },
    {
      "epoch": 0.08931690377199532,
      "grad_norm": 0.05280815809965134,
      "learning_rate": 8.738300808386935e-05,
      "loss": 10.2814,
      "step": 815
    },
    {
      "epoch": 0.08942649506496708,
      "grad_norm": 0.07005878537893295,
      "learning_rate": 8.717387607756713e-05,
      "loss": 10.2783,
      "step": 816
    },
    {
      "epoch": 0.08953608635793886,
      "grad_norm": 0.04553728178143501,
      "learning_rate": 8.696480109064342e-05,
      "loss": 10.288,
      "step": 817
    },
    {
      "epoch": 0.08964567765091064,
      "grad_norm": 0.05462590605020523,
      "learning_rate": 8.675578405255485e-05,
      "loss": 10.2698,
      "step": 818
    },
    {
      "epoch": 0.0897552689438824,
      "grad_norm": 0.05169070512056351,
      "learning_rate": 8.654682589250038e-05,
      "loss": 10.2823,
      "step": 819
    },
    {
      "epoch": 0.08986486023685418,
      "grad_norm": 0.049217816442251205,
      "learning_rate": 8.633792753941733e-05,
      "loss": 10.2756,
      "step": 820
    },
    {
      "epoch": 0.08997445152982596,
      "grad_norm": 0.04458007588982582,
      "learning_rate": 8.612908992197705e-05,
      "loss": 10.2685,
      "step": 821
    },
    {
      "epoch": 0.09008404282279772,
      "grad_norm": 0.056652605533599854,
      "learning_rate": 8.592031396858093e-05,
      "loss": 10.274,
      "step": 822
    },
    {
      "epoch": 0.0901936341157695,
      "grad_norm": 0.041533421725034714,
      "learning_rate": 8.571160060735624e-05,
      "loss": 10.2698,
      "step": 823
    },
    {
      "epoch": 0.09030322540874128,
      "grad_norm": 0.04980679973959923,
      "learning_rate": 8.550295076615188e-05,
      "loss": 10.2791,
      "step": 824
    },
    {
      "epoch": 0.09041281670171306,
      "grad_norm": 0.04616708308458328,
      "learning_rate": 8.529436537253458e-05,
      "loss": 10.271,
      "step": 825
    },
    {
      "epoch": 0.09052240799468482,
      "grad_norm": 0.05014638602733612,
      "learning_rate": 8.508584535378439e-05,
      "loss": 10.2641,
      "step": 826
    },
    {
      "epoch": 0.0906319992876566,
      "grad_norm": 0.05209904909133911,
      "learning_rate": 8.487739163689079e-05,
      "loss": 10.2689,
      "step": 827
    },
    {
      "epoch": 0.09074159058062838,
      "grad_norm": 0.04919566214084625,
      "learning_rate": 8.466900514854847e-05,
      "loss": 10.2688,
      "step": 828
    },
    {
      "epoch": 0.09085118187360014,
      "grad_norm": 0.05337287858128548,
      "learning_rate": 8.446068681515334e-05,
      "loss": 10.2774,
      "step": 829
    },
    {
      "epoch": 0.09096077316657192,
      "grad_norm": 0.05535747855901718,
      "learning_rate": 8.425243756279824e-05,
      "loss": 10.2774,
      "step": 830
    },
    {
      "epoch": 0.0910703644595437,
      "grad_norm": 0.05367989093065262,
      "learning_rate": 8.404425831726894e-05,
      "loss": 10.2783,
      "step": 831
    },
    {
      "epoch": 0.09117995575251546,
      "grad_norm": 0.04962741583585739,
      "learning_rate": 8.383615000404e-05,
      "loss": 10.2712,
      "step": 832
    },
    {
      "epoch": 0.09128954704548724,
      "grad_norm": 0.057855766266584396,
      "learning_rate": 8.362811354827059e-05,
      "loss": 10.2794,
      "step": 833
    },
    {
      "epoch": 0.09139913833845902,
      "grad_norm": 0.05004199594259262,
      "learning_rate": 8.342014987480047e-05,
      "loss": 10.2697,
      "step": 834
    },
    {
      "epoch": 0.09150872963143078,
      "grad_norm": 0.04604487866163254,
      "learning_rate": 8.321225990814591e-05,
      "loss": 10.2753,
      "step": 835
    },
    {
      "epoch": 0.09161832092440256,
      "grad_norm": 0.059143226593732834,
      "learning_rate": 8.300444457249543e-05,
      "loss": 10.268,
      "step": 836
    },
    {
      "epoch": 0.09172791221737434,
      "grad_norm": 0.05757003277540207,
      "learning_rate": 8.279670479170573e-05,
      "loss": 10.269,
      "step": 837
    },
    {
      "epoch": 0.0918375035103461,
      "grad_norm": 0.05382334068417549,
      "learning_rate": 8.258904148929775e-05,
      "loss": 10.2708,
      "step": 838
    },
    {
      "epoch": 0.09194709480331788,
      "grad_norm": 0.05317310243844986,
      "learning_rate": 8.238145558845235e-05,
      "loss": 10.2834,
      "step": 839
    },
    {
      "epoch": 0.09205668609628966,
      "grad_norm": 0.05128980427980423,
      "learning_rate": 8.217394801200631e-05,
      "loss": 10.2787,
      "step": 840
    },
    {
      "epoch": 0.09216627738926142,
      "grad_norm": 0.06840792298316956,
      "learning_rate": 8.196651968244826e-05,
      "loss": 10.2651,
      "step": 841
    },
    {
      "epoch": 0.0922758686822332,
      "grad_norm": 0.05317739397287369,
      "learning_rate": 8.175917152191447e-05,
      "loss": 10.2732,
      "step": 842
    },
    {
      "epoch": 0.09238545997520498,
      "grad_norm": 0.049440205097198486,
      "learning_rate": 8.15519044521848e-05,
      "loss": 10.2689,
      "step": 843
    },
    {
      "epoch": 0.09249505126817674,
      "grad_norm": 0.05720844864845276,
      "learning_rate": 8.134471939467874e-05,
      "loss": 10.2772,
      "step": 844
    },
    {
      "epoch": 0.09260464256114852,
      "grad_norm": 0.05964914709329605,
      "learning_rate": 8.113761727045105e-05,
      "loss": 10.2787,
      "step": 845
    },
    {
      "epoch": 0.0927142338541203,
      "grad_norm": 0.05455205589532852,
      "learning_rate": 8.093059900018792e-05,
      "loss": 10.2759,
      "step": 846
    },
    {
      "epoch": 0.09282382514709206,
      "grad_norm": 0.05345245450735092,
      "learning_rate": 8.072366550420266e-05,
      "loss": 10.2766,
      "step": 847
    },
    {
      "epoch": 0.09293341644006384,
      "grad_norm": 0.0705944076180458,
      "learning_rate": 8.051681770243175e-05,
      "loss": 10.282,
      "step": 848
    },
    {
      "epoch": 0.09304300773303562,
      "grad_norm": 0.0588788166642189,
      "learning_rate": 8.031005651443073e-05,
      "loss": 10.2618,
      "step": 849
    },
    {
      "epoch": 0.09315259902600738,
      "grad_norm": 0.0693337544798851,
      "learning_rate": 8.010338285937006e-05,
      "loss": 10.289,
      "step": 850
    },
    {
      "epoch": 0.09326219031897916,
      "grad_norm": 0.049039117991924286,
      "learning_rate": 7.989679765603108e-05,
      "loss": 10.2879,
      "step": 851
    },
    {
      "epoch": 0.09337178161195094,
      "grad_norm": 0.05503857135772705,
      "learning_rate": 7.969030182280192e-05,
      "loss": 10.2842,
      "step": 852
    },
    {
      "epoch": 0.0934813729049227,
      "grad_norm": 0.05541614070534706,
      "learning_rate": 7.948389627767343e-05,
      "loss": 10.2782,
      "step": 853
    },
    {
      "epoch": 0.09359096419789448,
      "grad_norm": 0.04641147702932358,
      "learning_rate": 7.927758193823501e-05,
      "loss": 10.2814,
      "step": 854
    },
    {
      "epoch": 0.09370055549086626,
      "grad_norm": 0.056134629994630814,
      "learning_rate": 7.907135972167069e-05,
      "loss": 10.2786,
      "step": 855
    },
    {
      "epoch": 0.09381014678383802,
      "grad_norm": 0.055640239268541336,
      "learning_rate": 7.88652305447549e-05,
      "loss": 10.2755,
      "step": 856
    },
    {
      "epoch": 0.0939197380768098,
      "grad_norm": 0.047932859510183334,
      "learning_rate": 7.865919532384844e-05,
      "loss": 10.2734,
      "step": 857
    },
    {
      "epoch": 0.09402932936978158,
      "grad_norm": 0.051452744752168655,
      "learning_rate": 7.845325497489449e-05,
      "loss": 10.2779,
      "step": 858
    },
    {
      "epoch": 0.09413892066275334,
      "grad_norm": 0.04715878516435623,
      "learning_rate": 7.82474104134144e-05,
      "loss": 10.276,
      "step": 859
    },
    {
      "epoch": 0.09424851195572512,
      "grad_norm": 0.04880620539188385,
      "learning_rate": 7.804166255450373e-05,
      "loss": 10.2817,
      "step": 860
    },
    {
      "epoch": 0.0943581032486969,
      "grad_norm": 0.05488169565796852,
      "learning_rate": 7.783601231282812e-05,
      "loss": 10.2698,
      "step": 861
    },
    {
      "epoch": 0.09446769454166866,
      "grad_norm": 0.05484771728515625,
      "learning_rate": 7.763046060261932e-05,
      "loss": 10.2766,
      "step": 862
    },
    {
      "epoch": 0.09457728583464044,
      "grad_norm": 0.060150861740112305,
      "learning_rate": 7.742500833767094e-05,
      "loss": 10.2783,
      "step": 863
    },
    {
      "epoch": 0.09468687712761222,
      "grad_norm": 0.0592849925160408,
      "learning_rate": 7.721965643133458e-05,
      "loss": 10.2861,
      "step": 864
    },
    {
      "epoch": 0.09479646842058398,
      "grad_norm": 0.06948515772819519,
      "learning_rate": 7.701440579651566e-05,
      "loss": 10.2706,
      "step": 865
    },
    {
      "epoch": 0.09490605971355576,
      "grad_norm": 0.04839586466550827,
      "learning_rate": 7.680925734566937e-05,
      "loss": 10.2796,
      "step": 866
    },
    {
      "epoch": 0.09501565100652754,
      "grad_norm": 0.04528702050447464,
      "learning_rate": 7.660421199079669e-05,
      "loss": 10.273,
      "step": 867
    },
    {
      "epoch": 0.0951252422994993,
      "grad_norm": 0.04802519828081131,
      "learning_rate": 7.639927064344022e-05,
      "loss": 10.2852,
      "step": 868
    },
    {
      "epoch": 0.09523483359247108,
      "grad_norm": 0.04629800096154213,
      "learning_rate": 7.619443421468021e-05,
      "loss": 10.2746,
      "step": 869
    },
    {
      "epoch": 0.09534442488544285,
      "grad_norm": 0.05719722434878349,
      "learning_rate": 7.598970361513051e-05,
      "loss": 10.2813,
      "step": 870
    },
    {
      "epoch": 0.09545401617841462,
      "grad_norm": 0.048237238079309464,
      "learning_rate": 7.578507975493448e-05,
      "loss": 10.2704,
      "step": 871
    },
    {
      "epoch": 0.0955636074713864,
      "grad_norm": 0.045861661434173584,
      "learning_rate": 7.558056354376098e-05,
      "loss": 10.2744,
      "step": 872
    },
    {
      "epoch": 0.09567319876435817,
      "grad_norm": 0.045138921588659286,
      "learning_rate": 7.537615589080027e-05,
      "loss": 10.2719,
      "step": 873
    },
    {
      "epoch": 0.09578279005732994,
      "grad_norm": 0.05323336273431778,
      "learning_rate": 7.517185770476006e-05,
      "loss": 10.2722,
      "step": 874
    },
    {
      "epoch": 0.09589238135030172,
      "grad_norm": 0.04510270059108734,
      "learning_rate": 7.496766989386136e-05,
      "loss": 10.2846,
      "step": 875
    },
    {
      "epoch": 0.0960019726432735,
      "grad_norm": 0.04218146577477455,
      "learning_rate": 7.476359336583454e-05,
      "loss": 10.275,
      "step": 876
    },
    {
      "epoch": 0.09611156393624526,
      "grad_norm": 0.04835299402475357,
      "learning_rate": 7.455962902791522e-05,
      "loss": 10.2698,
      "step": 877
    },
    {
      "epoch": 0.09622115522921704,
      "grad_norm": 0.05007461458444595,
      "learning_rate": 7.435577778684033e-05,
      "loss": 10.2769,
      "step": 878
    },
    {
      "epoch": 0.09633074652218881,
      "grad_norm": 0.05022592097520828,
      "learning_rate": 7.415204054884399e-05,
      "loss": 10.2673,
      "step": 879
    },
    {
      "epoch": 0.09644033781516058,
      "grad_norm": 0.049565937370061874,
      "learning_rate": 7.394841821965345e-05,
      "loss": 10.2703,
      "step": 880
    },
    {
      "epoch": 0.09654992910813236,
      "grad_norm": 0.04306262731552124,
      "learning_rate": 7.374491170448525e-05,
      "loss": 10.2838,
      "step": 881
    },
    {
      "epoch": 0.09665952040110413,
      "grad_norm": 0.054201725870370865,
      "learning_rate": 7.3541521908041e-05,
      "loss": 10.2692,
      "step": 882
    },
    {
      "epoch": 0.0967691116940759,
      "grad_norm": 0.05176425352692604,
      "learning_rate": 7.33382497345034e-05,
      "loss": 10.273,
      "step": 883
    },
    {
      "epoch": 0.09687870298704768,
      "grad_norm": 0.037117812782526016,
      "learning_rate": 7.313509608753231e-05,
      "loss": 10.2757,
      "step": 884
    },
    {
      "epoch": 0.09698829428001945,
      "grad_norm": 0.05107493698596954,
      "learning_rate": 7.293206187026061e-05,
      "loss": 10.2772,
      "step": 885
    },
    {
      "epoch": 0.09709788557299123,
      "grad_norm": 0.0571834035217762,
      "learning_rate": 7.27291479852903e-05,
      "loss": 10.2722,
      "step": 886
    },
    {
      "epoch": 0.097207476865963,
      "grad_norm": 0.05382966995239258,
      "learning_rate": 7.252635533468843e-05,
      "loss": 10.275,
      "step": 887
    },
    {
      "epoch": 0.09731706815893477,
      "grad_norm": 0.04790530353784561,
      "learning_rate": 7.232368481998309e-05,
      "loss": 10.2761,
      "step": 888
    },
    {
      "epoch": 0.09742665945190655,
      "grad_norm": 0.0543891005218029,
      "learning_rate": 7.212113734215932e-05,
      "loss": 10.2755,
      "step": 889
    },
    {
      "epoch": 0.09753625074487832,
      "grad_norm": 0.048300664871931076,
      "learning_rate": 7.191871380165538e-05,
      "loss": 10.2719,
      "step": 890
    },
    {
      "epoch": 0.0976458420378501,
      "grad_norm": 0.050636328756809235,
      "learning_rate": 7.17164150983584e-05,
      "loss": 10.2685,
      "step": 891
    },
    {
      "epoch": 0.09775543333082187,
      "grad_norm": 0.050630610436201096,
      "learning_rate": 7.151424213160061e-05,
      "loss": 10.2752,
      "step": 892
    },
    {
      "epoch": 0.09786502462379364,
      "grad_norm": 0.04724770784378052,
      "learning_rate": 7.131219580015521e-05,
      "loss": 10.2746,
      "step": 893
    },
    {
      "epoch": 0.09797461591676541,
      "grad_norm": 0.04564807564020157,
      "learning_rate": 7.11102770022325e-05,
      "loss": 10.2727,
      "step": 894
    },
    {
      "epoch": 0.09808420720973719,
      "grad_norm": 0.05080398917198181,
      "learning_rate": 7.090848663547574e-05,
      "loss": 10.2761,
      "step": 895
    },
    {
      "epoch": 0.09819379850270896,
      "grad_norm": 0.04780583828687668,
      "learning_rate": 7.070682559695736e-05,
      "loss": 10.2798,
      "step": 896
    },
    {
      "epoch": 0.09830338979568073,
      "grad_norm": 0.0631565973162651,
      "learning_rate": 7.050529478317476e-05,
      "loss": 10.2714,
      "step": 897
    },
    {
      "epoch": 0.09841298108865251,
      "grad_norm": 0.05332290753722191,
      "learning_rate": 7.03038950900464e-05,
      "loss": 10.2856,
      "step": 898
    },
    {
      "epoch": 0.09852257238162428,
      "grad_norm": 0.05657022446393967,
      "learning_rate": 7.010262741290798e-05,
      "loss": 10.2708,
      "step": 899
    },
    {
      "epoch": 0.09863216367459605,
      "grad_norm": 0.0584818497300148,
      "learning_rate": 6.990149264650814e-05,
      "loss": 10.2677,
      "step": 900
    },
    {
      "epoch": 0.09874175496756783,
      "grad_norm": 0.04040730372071266,
      "learning_rate": 6.970049168500474e-05,
      "loss": 10.2754,
      "step": 901
    },
    {
      "epoch": 0.0988513462605396,
      "grad_norm": 0.0403069332242012,
      "learning_rate": 6.94996254219608e-05,
      "loss": 10.2767,
      "step": 902
    },
    {
      "epoch": 0.09896093755351137,
      "grad_norm": 0.04640810191631317,
      "learning_rate": 6.929889475034048e-05,
      "loss": 10.2804,
      "step": 903
    },
    {
      "epoch": 0.09907052884648315,
      "grad_norm": 0.03677041828632355,
      "learning_rate": 6.909830056250527e-05,
      "loss": 10.2824,
      "step": 904
    },
    {
      "epoch": 0.09918012013945492,
      "grad_norm": 0.04863841086626053,
      "learning_rate": 6.889784375020978e-05,
      "loss": 10.2858,
      "step": 905
    },
    {
      "epoch": 0.0992897114324267,
      "grad_norm": 0.04434431344270706,
      "learning_rate": 6.869752520459803e-05,
      "loss": 10.2833,
      "step": 906
    },
    {
      "epoch": 0.09939930272539847,
      "grad_norm": 0.05715576186776161,
      "learning_rate": 6.849734581619918e-05,
      "loss": 10.277,
      "step": 907
    },
    {
      "epoch": 0.09950889401837024,
      "grad_norm": 0.05619329959154129,
      "learning_rate": 6.829730647492404e-05,
      "loss": 10.2851,
      "step": 908
    },
    {
      "epoch": 0.09961848531134201,
      "grad_norm": 0.052352823317050934,
      "learning_rate": 6.80974080700606e-05,
      "loss": 10.2812,
      "step": 909
    },
    {
      "epoch": 0.09972807660431379,
      "grad_norm": 0.048739638179540634,
      "learning_rate": 6.789765149027039e-05,
      "loss": 10.2797,
      "step": 910
    },
    {
      "epoch": 0.09983766789728556,
      "grad_norm": 0.04787062853574753,
      "learning_rate": 6.769803762358443e-05,
      "loss": 10.273,
      "step": 911
    },
    {
      "epoch": 0.09994725919025733,
      "grad_norm": 0.04848838225007057,
      "learning_rate": 6.749856735739928e-05,
      "loss": 10.2731,
      "step": 912
    },
    {
      "epoch": 0.10005685048322911,
      "grad_norm": 0.054745253175497055,
      "learning_rate": 6.729924157847323e-05,
      "loss": 10.2757,
      "step": 913
    },
    {
      "epoch": 0.10016644177620088,
      "grad_norm": 0.05668586492538452,
      "learning_rate": 6.710006117292209e-05,
      "loss": 10.2862,
      "step": 914
    },
    {
      "epoch": 0.10027603306917265,
      "grad_norm": 0.052677858620882034,
      "learning_rate": 6.690102702621548e-05,
      "loss": 10.2716,
      "step": 915
    },
    {
      "epoch": 0.10038562436214443,
      "grad_norm": 0.053779006004333496,
      "learning_rate": 6.670214002317278e-05,
      "loss": 10.2813,
      "step": 916
    },
    {
      "epoch": 0.1004952156551162,
      "grad_norm": 0.04740731045603752,
      "learning_rate": 6.650340104795932e-05,
      "loss": 10.2759,
      "step": 917
    },
    {
      "epoch": 0.10060480694808797,
      "grad_norm": 0.04593643546104431,
      "learning_rate": 6.630481098408228e-05,
      "loss": 10.2763,
      "step": 918
    },
    {
      "epoch": 0.10071439824105975,
      "grad_norm": 0.04646601900458336,
      "learning_rate": 6.610637071438686e-05,
      "loss": 10.281,
      "step": 919
    },
    {
      "epoch": 0.10082398953403152,
      "grad_norm": 0.05093095451593399,
      "learning_rate": 6.590808112105232e-05,
      "loss": 10.2743,
      "step": 920
    },
    {
      "epoch": 0.1009335808270033,
      "grad_norm": 0.04974231496453285,
      "learning_rate": 6.570994308558812e-05,
      "loss": 10.2775,
      "step": 921
    },
    {
      "epoch": 0.10104317211997507,
      "grad_norm": 0.04603124409914017,
      "learning_rate": 6.551195748882997e-05,
      "loss": 10.27,
      "step": 922
    },
    {
      "epoch": 0.10115276341294684,
      "grad_norm": 0.039514947682619095,
      "learning_rate": 6.531412521093586e-05,
      "loss": 10.2725,
      "step": 923
    },
    {
      "epoch": 0.10126235470591861,
      "grad_norm": 0.04194781556725502,
      "learning_rate": 6.51164471313822e-05,
      "loss": 10.2723,
      "step": 924
    },
    {
      "epoch": 0.10137194599889039,
      "grad_norm": 0.047541651874780655,
      "learning_rate": 6.491892412895995e-05,
      "loss": 10.2776,
      "step": 925
    },
    {
      "epoch": 0.10148153729186216,
      "grad_norm": 0.043093204498291016,
      "learning_rate": 6.472155708177052e-05,
      "loss": 10.2739,
      "step": 926
    },
    {
      "epoch": 0.10159112858483393,
      "grad_norm": 0.04551133140921593,
      "learning_rate": 6.452434686722224e-05,
      "loss": 10.2772,
      "step": 927
    },
    {
      "epoch": 0.10170071987780571,
      "grad_norm": 0.05120580643415451,
      "learning_rate": 6.432729436202604e-05,
      "loss": 10.2724,
      "step": 928
    },
    {
      "epoch": 0.10181031117077748,
      "grad_norm": 0.048426371067762375,
      "learning_rate": 6.41304004421918e-05,
      "loss": 10.2774,
      "step": 929
    },
    {
      "epoch": 0.10191990246374925,
      "grad_norm": 0.049765702337026596,
      "learning_rate": 6.393366598302446e-05,
      "loss": 10.2728,
      "step": 930
    },
    {
      "epoch": 0.10202949375672103,
      "grad_norm": 0.051022469997406006,
      "learning_rate": 6.373709185911998e-05,
      "loss": 10.2768,
      "step": 931
    },
    {
      "epoch": 0.1021390850496928,
      "grad_norm": 0.05051299184560776,
      "learning_rate": 6.354067894436155e-05,
      "loss": 10.2664,
      "step": 932
    },
    {
      "epoch": 0.10224867634266457,
      "grad_norm": 0.052603740245103836,
      "learning_rate": 6.334442811191576e-05,
      "loss": 10.2688,
      "step": 933
    },
    {
      "epoch": 0.10235826763563635,
      "grad_norm": 0.056494902819395065,
      "learning_rate": 6.314834023422858e-05,
      "loss": 10.2723,
      "step": 934
    },
    {
      "epoch": 0.10246785892860812,
      "grad_norm": 0.047464076429605484,
      "learning_rate": 6.295241618302156e-05,
      "loss": 10.2798,
      "step": 935
    },
    {
      "epoch": 0.1025774502215799,
      "grad_norm": 0.06030727177858353,
      "learning_rate": 6.275665682928803e-05,
      "loss": 10.2721,
      "step": 936
    },
    {
      "epoch": 0.10268704151455167,
      "grad_norm": 0.05031837895512581,
      "learning_rate": 6.256106304328905e-05,
      "loss": 10.2677,
      "step": 937
    },
    {
      "epoch": 0.10279663280752344,
      "grad_norm": 0.06392553448677063,
      "learning_rate": 6.23656356945497e-05,
      "loss": 10.274,
      "step": 938
    },
    {
      "epoch": 0.10290622410049521,
      "grad_norm": 0.052959784865379333,
      "learning_rate": 6.21703756518551e-05,
      "loss": 10.2699,
      "step": 939
    },
    {
      "epoch": 0.10301581539346699,
      "grad_norm": 0.0578688345849514,
      "learning_rate": 6.197528378324665e-05,
      "loss": 10.2832,
      "step": 940
    },
    {
      "epoch": 0.10312540668643876,
      "grad_norm": 0.0566239207983017,
      "learning_rate": 6.17803609560181e-05,
      "loss": 10.2733,
      "step": 941
    },
    {
      "epoch": 0.10323499797941053,
      "grad_norm": 0.05225759372115135,
      "learning_rate": 6.158560803671168e-05,
      "loss": 10.2642,
      "step": 942
    },
    {
      "epoch": 0.10334458927238231,
      "grad_norm": 0.05896095931529999,
      "learning_rate": 6.139102589111435e-05,
      "loss": 10.2711,
      "step": 943
    },
    {
      "epoch": 0.10345418056535408,
      "grad_norm": 0.05416179075837135,
      "learning_rate": 6.119661538425381e-05,
      "loss": 10.2736,
      "step": 944
    },
    {
      "epoch": 0.10356377185832585,
      "grad_norm": 0.058569565415382385,
      "learning_rate": 6.100237738039484e-05,
      "loss": 10.2716,
      "step": 945
    },
    {
      "epoch": 0.10367336315129763,
      "grad_norm": 0.05686130374670029,
      "learning_rate": 6.0808312743035236e-05,
      "loss": 10.2643,
      "step": 946
    },
    {
      "epoch": 0.10378295444426941,
      "grad_norm": 0.05203365534543991,
      "learning_rate": 6.061442233490211e-05,
      "loss": 10.2721,
      "step": 947
    },
    {
      "epoch": 0.10389254573724117,
      "grad_norm": 0.06044309586286545,
      "learning_rate": 6.042070701794806e-05,
      "loss": 10.2709,
      "step": 948
    },
    {
      "epoch": 0.10400213703021295,
      "grad_norm": 0.06922642141580582,
      "learning_rate": 6.0227167653347305e-05,
      "loss": 10.2746,
      "step": 949
    },
    {
      "epoch": 0.10411172832318473,
      "grad_norm": 0.07655718177556992,
      "learning_rate": 6.0033805101491794e-05,
      "loss": 10.2696,
      "step": 950
    },
    {
      "epoch": 0.1042213196161565,
      "grad_norm": 0.04576713591814041,
      "learning_rate": 5.98406202219875e-05,
      "loss": 10.278,
      "step": 951
    },
    {
      "epoch": 0.10433091090912827,
      "grad_norm": 0.03485884517431259,
      "learning_rate": 5.964761387365052e-05,
      "loss": 10.285,
      "step": 952
    },
    {
      "epoch": 0.10444050220210005,
      "grad_norm": 0.04001082479953766,
      "learning_rate": 5.9454786914503255e-05,
      "loss": 10.2826,
      "step": 953
    },
    {
      "epoch": 0.10455009349507181,
      "grad_norm": 0.05416722595691681,
      "learning_rate": 5.926214020177074e-05,
      "loss": 10.2712,
      "step": 954
    },
    {
      "epoch": 0.10465968478804359,
      "grad_norm": 0.0446426160633564,
      "learning_rate": 5.9069674591876534e-05,
      "loss": 10.2783,
      "step": 955
    },
    {
      "epoch": 0.10476927608101537,
      "grad_norm": 0.056975703686475754,
      "learning_rate": 5.887739094043923e-05,
      "loss": 10.2836,
      "step": 956
    },
    {
      "epoch": 0.10487886737398713,
      "grad_norm": 0.045389216393232346,
      "learning_rate": 5.868529010226845e-05,
      "loss": 10.2809,
      "step": 957
    },
    {
      "epoch": 0.10498845866695891,
      "grad_norm": 0.06334411352872849,
      "learning_rate": 5.849337293136112e-05,
      "loss": 10.2838,
      "step": 958
    },
    {
      "epoch": 0.10509804995993069,
      "grad_norm": 0.06192444637417793,
      "learning_rate": 5.830164028089766e-05,
      "loss": 10.2853,
      "step": 959
    },
    {
      "epoch": 0.10520764125290245,
      "grad_norm": 0.0474047027528286,
      "learning_rate": 5.811009300323818e-05,
      "loss": 10.2768,
      "step": 960
    },
    {
      "epoch": 0.10531723254587423,
      "grad_norm": 0.05903308466076851,
      "learning_rate": 5.791873194991872e-05,
      "loss": 10.2776,
      "step": 961
    },
    {
      "epoch": 0.10542682383884601,
      "grad_norm": 0.0565553680062294,
      "learning_rate": 5.7727557971647427e-05,
      "loss": 10.2846,
      "step": 962
    },
    {
      "epoch": 0.10553641513181777,
      "grad_norm": 0.0588684119284153,
      "learning_rate": 5.7536571918300864e-05,
      "loss": 10.276,
      "step": 963
    },
    {
      "epoch": 0.10564600642478955,
      "grad_norm": 0.06081688776612282,
      "learning_rate": 5.734577463892008e-05,
      "loss": 10.2818,
      "step": 964
    },
    {
      "epoch": 0.10575559771776133,
      "grad_norm": 0.06557998061180115,
      "learning_rate": 5.7155166981706956e-05,
      "loss": 10.2903,
      "step": 965
    },
    {
      "epoch": 0.1058651890107331,
      "grad_norm": 0.060764044523239136,
      "learning_rate": 5.6964749794020354e-05,
      "loss": 10.2819,
      "step": 966
    },
    {
      "epoch": 0.10597478030370487,
      "grad_norm": 0.05271344631910324,
      "learning_rate": 5.6774523922372394e-05,
      "loss": 10.2702,
      "step": 967
    },
    {
      "epoch": 0.10608437159667665,
      "grad_norm": 0.05223078653216362,
      "learning_rate": 5.6584490212424804e-05,
      "loss": 10.2728,
      "step": 968
    },
    {
      "epoch": 0.10619396288964841,
      "grad_norm": 0.04828125238418579,
      "learning_rate": 5.639464950898491e-05,
      "loss": 10.2804,
      "step": 969
    },
    {
      "epoch": 0.10630355418262019,
      "grad_norm": 0.04825196415185928,
      "learning_rate": 5.620500265600206e-05,
      "loss": 10.2785,
      "step": 970
    },
    {
      "epoch": 0.10641314547559197,
      "grad_norm": 0.050081104040145874,
      "learning_rate": 5.601555049656382e-05,
      "loss": 10.2676,
      "step": 971
    },
    {
      "epoch": 0.10652273676856373,
      "grad_norm": 0.04407525435090065,
      "learning_rate": 5.58262938728922e-05,
      "loss": 10.2669,
      "step": 972
    },
    {
      "epoch": 0.10663232806153551,
      "grad_norm": 0.052682019770145416,
      "learning_rate": 5.563723362634008e-05,
      "loss": 10.2757,
      "step": 973
    },
    {
      "epoch": 0.10674191935450729,
      "grad_norm": 0.04584966599941254,
      "learning_rate": 5.544837059738719e-05,
      "loss": 10.2692,
      "step": 974
    },
    {
      "epoch": 0.10685151064747905,
      "grad_norm": 0.049827199429273605,
      "learning_rate": 5.525970562563656e-05,
      "loss": 10.2762,
      "step": 975
    },
    {
      "epoch": 0.10696110194045083,
      "grad_norm": 0.05259262025356293,
      "learning_rate": 5.507123954981073e-05,
      "loss": 10.2678,
      "step": 976
    },
    {
      "epoch": 0.10707069323342261,
      "grad_norm": 0.05048719793558121,
      "learning_rate": 5.488297320774807e-05,
      "loss": 10.2751,
      "step": 977
    },
    {
      "epoch": 0.10718028452639437,
      "grad_norm": 0.04920447617769241,
      "learning_rate": 5.4694907436399e-05,
      "loss": 10.2762,
      "step": 978
    },
    {
      "epoch": 0.10728987581936615,
      "grad_norm": 0.053856030106544495,
      "learning_rate": 5.4507043071822284e-05,
      "loss": 10.2828,
      "step": 979
    },
    {
      "epoch": 0.10739946711233793,
      "grad_norm": 0.04575687274336815,
      "learning_rate": 5.431938094918132e-05,
      "loss": 10.2724,
      "step": 980
    },
    {
      "epoch": 0.1075090584053097,
      "grad_norm": 0.0615171454846859,
      "learning_rate": 5.41319219027404e-05,
      "loss": 10.2737,
      "step": 981
    },
    {
      "epoch": 0.10761864969828147,
      "grad_norm": 0.04373757541179657,
      "learning_rate": 5.394466676586114e-05,
      "loss": 10.2776,
      "step": 982
    },
    {
      "epoch": 0.10772824099125325,
      "grad_norm": 0.057647306472063065,
      "learning_rate": 5.375761637099854e-05,
      "loss": 10.2836,
      "step": 983
    },
    {
      "epoch": 0.10783783228422501,
      "grad_norm": 0.04217182472348213,
      "learning_rate": 5.357077154969742e-05,
      "loss": 10.2695,
      "step": 984
    },
    {
      "epoch": 0.10794742357719679,
      "grad_norm": 0.053424444049596786,
      "learning_rate": 5.3384133132588784e-05,
      "loss": 10.2635,
      "step": 985
    },
    {
      "epoch": 0.10805701487016857,
      "grad_norm": 0.04964132234454155,
      "learning_rate": 5.3197701949386e-05,
      "loss": 10.2611,
      "step": 986
    },
    {
      "epoch": 0.10816660616314033,
      "grad_norm": 0.05222456529736519,
      "learning_rate": 5.301147882888116e-05,
      "loss": 10.2668,
      "step": 987
    },
    {
      "epoch": 0.10827619745611211,
      "grad_norm": 0.04807818681001663,
      "learning_rate": 5.28254645989414e-05,
      "loss": 10.2699,
      "step": 988
    },
    {
      "epoch": 0.10838578874908389,
      "grad_norm": 0.0519399419426918,
      "learning_rate": 5.2639660086505226e-05,
      "loss": 10.2665,
      "step": 989
    },
    {
      "epoch": 0.10849538004205565,
      "grad_norm": 0.04990521818399429,
      "learning_rate": 5.2454066117578815e-05,
      "loss": 10.2784,
      "step": 990
    },
    {
      "epoch": 0.10860497133502743,
      "grad_norm": 0.05557786300778389,
      "learning_rate": 5.226868351723244e-05,
      "loss": 10.2717,
      "step": 991
    },
    {
      "epoch": 0.10871456262799921,
      "grad_norm": 0.04949048161506653,
      "learning_rate": 5.2083513109596616e-05,
      "loss": 10.2665,
      "step": 992
    },
    {
      "epoch": 0.10882415392097097,
      "grad_norm": 0.053594693541526794,
      "learning_rate": 5.189855571785859e-05,
      "loss": 10.2702,
      "step": 993
    },
    {
      "epoch": 0.10893374521394275,
      "grad_norm": 0.054413311183452606,
      "learning_rate": 5.171381216425863e-05,
      "loss": 10.2763,
      "step": 994
    },
    {
      "epoch": 0.10904333650691453,
      "grad_norm": 0.05530654639005661,
      "learning_rate": 5.152928327008635e-05,
      "loss": 10.2722,
      "step": 995
    },
    {
      "epoch": 0.10915292779988629,
      "grad_norm": 0.05560198798775673,
      "learning_rate": 5.134496985567714e-05,
      "loss": 10.273,
      "step": 996
    },
    {
      "epoch": 0.10926251909285807,
      "grad_norm": 0.0545300655066967,
      "learning_rate": 5.116087274040837e-05,
      "loss": 10.2712,
      "step": 997
    },
    {
      "epoch": 0.10937211038582985,
      "grad_norm": 0.052571315318346024,
      "learning_rate": 5.0976992742695925e-05,
      "loss": 10.272,
      "step": 998
    },
    {
      "epoch": 0.10948170167880161,
      "grad_norm": 0.06803826242685318,
      "learning_rate": 5.07933306799904e-05,
      "loss": 10.2606,
      "step": 999
    },
    {
      "epoch": 0.10959129297177339,
      "grad_norm": 0.059746239334344864,
      "learning_rate": 5.060988736877366e-05,
      "loss": 10.2757,
      "step": 1000
    },
    {
      "epoch": 0.10970088426474517,
      "grad_norm": 0.03651243820786476,
      "learning_rate": 5.042666362455498e-05,
      "loss": 10.2796,
      "step": 1001
    },
    {
      "epoch": 0.10981047555771693,
      "grad_norm": 0.037787314504384995,
      "learning_rate": 5.024366026186755e-05,
      "loss": 10.2754,
      "step": 1002
    },
    {
      "epoch": 0.10992006685068871,
      "grad_norm": 0.03978487476706505,
      "learning_rate": 5.006087809426496e-05,
      "loss": 10.2822,
      "step": 1003
    },
    {
      "epoch": 0.11002965814366049,
      "grad_norm": 0.04848557710647583,
      "learning_rate": 4.987831793431731e-05,
      "loss": 10.2813,
      "step": 1004
    },
    {
      "epoch": 0.11013924943663225,
      "grad_norm": 0.045813895761966705,
      "learning_rate": 4.9695980593607817e-05,
      "loss": 10.2796,
      "step": 1005
    },
    {
      "epoch": 0.11024884072960403,
      "grad_norm": 0.0440620556473732,
      "learning_rate": 4.9513866882729146e-05,
      "loss": 10.2703,
      "step": 1006
    },
    {
      "epoch": 0.11035843202257581,
      "grad_norm": 0.042686086148023605,
      "learning_rate": 4.9331977611279777e-05,
      "loss": 10.2852,
      "step": 1007
    },
    {
      "epoch": 0.11046802331554759,
      "grad_norm": 0.047057051211595535,
      "learning_rate": 4.9150313587860433e-05,
      "loss": 10.2691,
      "step": 1008
    },
    {
      "epoch": 0.11057761460851935,
      "grad_norm": 0.06092926859855652,
      "learning_rate": 4.896887562007054e-05,
      "loss": 10.2779,
      "step": 1009
    },
    {
      "epoch": 0.11068720590149113,
      "grad_norm": 0.04406747221946716,
      "learning_rate": 4.8787664514504504e-05,
      "loss": 10.2866,
      "step": 1010
    },
    {
      "epoch": 0.1107967971944629,
      "grad_norm": 0.04836060479283333,
      "learning_rate": 4.860668107674823e-05,
      "loss": 10.2764,
      "step": 1011
    },
    {
      "epoch": 0.11090638848743467,
      "grad_norm": 0.05875762924551964,
      "learning_rate": 4.8425926111375506e-05,
      "loss": 10.2683,
      "step": 1012
    },
    {
      "epoch": 0.11101597978040645,
      "grad_norm": 0.0507938377559185,
      "learning_rate": 4.824540042194443e-05,
      "loss": 10.2868,
      "step": 1013
    },
    {
      "epoch": 0.11112557107337823,
      "grad_norm": 0.05853484570980072,
      "learning_rate": 4.8065104810993856e-05,
      "loss": 10.2854,
      "step": 1014
    },
    {
      "epoch": 0.11123516236634999,
      "grad_norm": 0.061149124056100845,
      "learning_rate": 4.788504008003978e-05,
      "loss": 10.266,
      "step": 1015
    },
    {
      "epoch": 0.11134475365932177,
      "grad_norm": 0.07246897369623184,
      "learning_rate": 4.770520702957182e-05,
      "loss": 10.2942,
      "step": 1016
    },
    {
      "epoch": 0.11145434495229355,
      "grad_norm": 0.04815657436847687,
      "learning_rate": 4.752560645904962e-05,
      "loss": 10.2688,
      "step": 1017
    },
    {
      "epoch": 0.11156393624526531,
      "grad_norm": 0.05019097775220871,
      "learning_rate": 4.734623916689941e-05,
      "loss": 10.2732,
      "step": 1018
    },
    {
      "epoch": 0.11167352753823709,
      "grad_norm": 0.05417206883430481,
      "learning_rate": 4.716710595051022e-05,
      "loss": 10.2838,
      "step": 1019
    },
    {
      "epoch": 0.11178311883120887,
      "grad_norm": 0.05732262507081032,
      "learning_rate": 4.698820760623064e-05,
      "loss": 10.277,
      "step": 1020
    },
    {
      "epoch": 0.11189271012418063,
      "grad_norm": 0.042488571256399155,
      "learning_rate": 4.6809544929365004e-05,
      "loss": 10.2728,
      "step": 1021
    },
    {
      "epoch": 0.11200230141715241,
      "grad_norm": 0.04481584206223488,
      "learning_rate": 4.663111871417e-05,
      "loss": 10.2729,
      "step": 1022
    },
    {
      "epoch": 0.11211189271012419,
      "grad_norm": 0.05379185080528259,
      "learning_rate": 4.645292975385111e-05,
      "loss": 10.276,
      "step": 1023
    },
    {
      "epoch": 0.11222148400309595,
      "grad_norm": 0.041503023356199265,
      "learning_rate": 4.627497884055912e-05,
      "loss": 10.2796,
      "step": 1024
    },
    {
      "epoch": 0.11233107529606773,
      "grad_norm": 0.04313621670007706,
      "learning_rate": 4.609726676538652e-05,
      "loss": 10.2761,
      "step": 1025
    },
    {
      "epoch": 0.1124406665890395,
      "grad_norm": 0.04391136020421982,
      "learning_rate": 4.591979431836402e-05,
      "loss": 10.2729,
      "step": 1026
    },
    {
      "epoch": 0.11255025788201127,
      "grad_norm": 0.04332420229911804,
      "learning_rate": 4.574256228845706e-05,
      "loss": 10.2766,
      "step": 1027
    },
    {
      "epoch": 0.11265984917498305,
      "grad_norm": 0.04936756193637848,
      "learning_rate": 4.5565571463562365e-05,
      "loss": 10.2825,
      "step": 1028
    },
    {
      "epoch": 0.11276944046795483,
      "grad_norm": 0.045753590762615204,
      "learning_rate": 4.5388822630504256e-05,
      "loss": 10.2657,
      "step": 1029
    },
    {
      "epoch": 0.11287903176092659,
      "grad_norm": 0.049652982503175735,
      "learning_rate": 4.521231657503132e-05,
      "loss": 10.2725,
      "step": 1030
    },
    {
      "epoch": 0.11298862305389837,
      "grad_norm": 0.04284120723605156,
      "learning_rate": 4.503605408181286e-05,
      "loss": 10.2726,
      "step": 1031
    },
    {
      "epoch": 0.11309821434687015,
      "grad_norm": 0.05872884392738342,
      "learning_rate": 4.486003593443537e-05,
      "loss": 10.2645,
      "step": 1032
    },
    {
      "epoch": 0.11320780563984191,
      "grad_norm": 0.047861963510513306,
      "learning_rate": 4.468426291539914e-05,
      "loss": 10.2732,
      "step": 1033
    },
    {
      "epoch": 0.11331739693281369,
      "grad_norm": 0.041527003049850464,
      "learning_rate": 4.4508735806114654e-05,
      "loss": 10.2742,
      "step": 1034
    },
    {
      "epoch": 0.11342698822578547,
      "grad_norm": 0.0502019077539444,
      "learning_rate": 4.433345538689929e-05,
      "loss": 10.2697,
      "step": 1035
    },
    {
      "epoch": 0.11353657951875723,
      "grad_norm": 0.041743937879800797,
      "learning_rate": 4.415842243697369e-05,
      "loss": 10.2793,
      "step": 1036
    },
    {
      "epoch": 0.11364617081172901,
      "grad_norm": 0.047201234847307205,
      "learning_rate": 4.39836377344583e-05,
      "loss": 10.2685,
      "step": 1037
    },
    {
      "epoch": 0.11375576210470079,
      "grad_norm": 0.046231336891651154,
      "learning_rate": 4.380910205637012e-05,
      "loss": 10.268,
      "step": 1038
    },
    {
      "epoch": 0.11386535339767255,
      "grad_norm": 0.05502534657716751,
      "learning_rate": 4.363481617861893e-05,
      "loss": 10.2711,
      "step": 1039
    },
    {
      "epoch": 0.11397494469064433,
      "grad_norm": 0.05457928776741028,
      "learning_rate": 4.346078087600412e-05,
      "loss": 10.2704,
      "step": 1040
    },
    {
      "epoch": 0.1140845359836161,
      "grad_norm": 0.06011061742901802,
      "learning_rate": 4.3286996922211034e-05,
      "loss": 10.2691,
      "step": 1041
    },
    {
      "epoch": 0.11419412727658787,
      "grad_norm": 0.05993199720978737,
      "learning_rate": 4.311346508980772e-05,
      "loss": 10.2743,
      "step": 1042
    },
    {
      "epoch": 0.11430371856955965,
      "grad_norm": 0.05745707452297211,
      "learning_rate": 4.2940186150241365e-05,
      "loss": 10.2696,
      "step": 1043
    },
    {
      "epoch": 0.11441330986253143,
      "grad_norm": 0.04836670681834221,
      "learning_rate": 4.27671608738349e-05,
      "loss": 10.2773,
      "step": 1044
    },
    {
      "epoch": 0.11452290115550319,
      "grad_norm": 0.04864399507641792,
      "learning_rate": 4.2594390029783534e-05,
      "loss": 10.2718,
      "step": 1045
    },
    {
      "epoch": 0.11463249244847497,
      "grad_norm": 0.057778771966695786,
      "learning_rate": 4.242187438615153e-05,
      "loss": 10.2777,
      "step": 1046
    },
    {
      "epoch": 0.11474208374144675,
      "grad_norm": 0.0638226866722107,
      "learning_rate": 4.224961470986849e-05,
      "loss": 10.2646,
      "step": 1047
    },
    {
      "epoch": 0.11485167503441851,
      "grad_norm": 0.06045162305235863,
      "learning_rate": 4.207761176672614e-05,
      "loss": 10.2676,
      "step": 1048
    },
    {
      "epoch": 0.11496126632739029,
      "grad_norm": 0.05830172076821327,
      "learning_rate": 4.190586632137491e-05,
      "loss": 10.2718,
      "step": 1049
    },
    {
      "epoch": 0.11507085762036207,
      "grad_norm": 0.06628182530403137,
      "learning_rate": 4.173437913732048e-05,
      "loss": 10.2738,
      "step": 1050
    },
    {
      "epoch": 0.11518044891333383,
      "grad_norm": 0.042743176221847534,
      "learning_rate": 4.156315097692037e-05,
      "loss": 10.281,
      "step": 1051
    },
    {
      "epoch": 0.11529004020630561,
      "grad_norm": 0.04471007362008095,
      "learning_rate": 4.139218260138074e-05,
      "loss": 10.2833,
      "step": 1052
    },
    {
      "epoch": 0.11539963149927739,
      "grad_norm": 0.0423806831240654,
      "learning_rate": 4.12214747707527e-05,
      "loss": 10.2691,
      "step": 1053
    },
    {
      "epoch": 0.11550922279224915,
      "grad_norm": 0.04028995707631111,
      "learning_rate": 4.1051028243929125e-05,
      "loss": 10.2699,
      "step": 1054
    },
    {
      "epoch": 0.11561881408522093,
      "grad_norm": 0.04998929426074028,
      "learning_rate": 4.088084377864135e-05,
      "loss": 10.2727,
      "step": 1055
    },
    {
      "epoch": 0.1157284053781927,
      "grad_norm": 0.04475776106119156,
      "learning_rate": 4.07109221314556e-05,
      "loss": 10.2796,
      "step": 1056
    },
    {
      "epoch": 0.11583799667116447,
      "grad_norm": 0.05487216264009476,
      "learning_rate": 4.054126405776971e-05,
      "loss": 10.2846,
      "step": 1057
    },
    {
      "epoch": 0.11594758796413625,
      "grad_norm": 0.04862336069345474,
      "learning_rate": 4.037187031180985e-05,
      "loss": 10.2766,
      "step": 1058
    },
    {
      "epoch": 0.11605717925710803,
      "grad_norm": 0.04893920198082924,
      "learning_rate": 4.020274164662707e-05,
      "loss": 10.2742,
      "step": 1059
    },
    {
      "epoch": 0.11616677055007979,
      "grad_norm": 0.04978443682193756,
      "learning_rate": 4.003387881409397e-05,
      "loss": 10.2794,
      "step": 1060
    },
    {
      "epoch": 0.11627636184305157,
      "grad_norm": 0.0513479970395565,
      "learning_rate": 3.986528256490141e-05,
      "loss": 10.2746,
      "step": 1061
    },
    {
      "epoch": 0.11638595313602335,
      "grad_norm": 0.050011079758405685,
      "learning_rate": 3.969695364855511e-05,
      "loss": 10.281,
      "step": 1062
    },
    {
      "epoch": 0.11649554442899511,
      "grad_norm": 0.05664727836847305,
      "learning_rate": 3.952889281337235e-05,
      "loss": 10.2762,
      "step": 1063
    },
    {
      "epoch": 0.11660513572196689,
      "grad_norm": 0.054064568132162094,
      "learning_rate": 3.93611008064786e-05,
      "loss": 10.2698,
      "step": 1064
    },
    {
      "epoch": 0.11671472701493867,
      "grad_norm": 0.053642019629478455,
      "learning_rate": 3.9193578373804364e-05,
      "loss": 10.2767,
      "step": 1065
    },
    {
      "epoch": 0.11682431830791043,
      "grad_norm": 0.05946362763643265,
      "learning_rate": 3.90263262600816e-05,
      "loss": 10.2829,
      "step": 1066
    },
    {
      "epoch": 0.11693390960088221,
      "grad_norm": 0.04508134350180626,
      "learning_rate": 3.88593452088406e-05,
      "loss": 10.2837,
      "step": 1067
    },
    {
      "epoch": 0.11704350089385399,
      "grad_norm": 0.03719077259302139,
      "learning_rate": 3.869263596240661e-05,
      "loss": 10.2797,
      "step": 1068
    },
    {
      "epoch": 0.11715309218682576,
      "grad_norm": 0.04885408282279968,
      "learning_rate": 3.8526199261896544e-05,
      "loss": 10.2688,
      "step": 1069
    },
    {
      "epoch": 0.11726268347979753,
      "grad_norm": 0.03643273934721947,
      "learning_rate": 3.836003584721577e-05,
      "loss": 10.2678,
      "step": 1070
    },
    {
      "epoch": 0.1173722747727693,
      "grad_norm": 0.041023410856723785,
      "learning_rate": 3.8194146457054655e-05,
      "loss": 10.2671,
      "step": 1071
    },
    {
      "epoch": 0.11748186606574108,
      "grad_norm": 0.039677999913692474,
      "learning_rate": 3.802853182888543e-05,
      "loss": 10.2775,
      "step": 1072
    },
    {
      "epoch": 0.11759145735871285,
      "grad_norm": 0.05191703885793686,
      "learning_rate": 3.786319269895877e-05,
      "loss": 10.2745,
      "step": 1073
    },
    {
      "epoch": 0.11770104865168463,
      "grad_norm": 0.04453219845890999,
      "learning_rate": 3.769812980230074e-05,
      "loss": 10.2794,
      "step": 1074
    },
    {
      "epoch": 0.1178106399446564,
      "grad_norm": 0.054414961487054825,
      "learning_rate": 3.7533343872709294e-05,
      "loss": 10.2773,
      "step": 1075
    },
    {
      "epoch": 0.11792023123762817,
      "grad_norm": 0.046708233654499054,
      "learning_rate": 3.736883564275112e-05,
      "loss": 10.2641,
      "step": 1076
    },
    {
      "epoch": 0.11802982253059995,
      "grad_norm": 0.03794122859835625,
      "learning_rate": 3.7204605843758386e-05,
      "loss": 10.2791,
      "step": 1077
    },
    {
      "epoch": 0.11813941382357172,
      "grad_norm": 0.057558685541152954,
      "learning_rate": 3.704065520582549e-05,
      "loss": 10.2746,
      "step": 1078
    },
    {
      "epoch": 0.11824900511654349,
      "grad_norm": 0.05685292184352875,
      "learning_rate": 3.6876984457805786e-05,
      "loss": 10.2711,
      "step": 1079
    },
    {
      "epoch": 0.11835859640951527,
      "grad_norm": 0.053439125418663025,
      "learning_rate": 3.671359432730834e-05,
      "loss": 10.2723,
      "step": 1080
    },
    {
      "epoch": 0.11846818770248704,
      "grad_norm": 0.049537088721990585,
      "learning_rate": 3.655048554069478e-05,
      "loss": 10.271,
      "step": 1081
    },
    {
      "epoch": 0.11857777899545881,
      "grad_norm": 0.056646306067705154,
      "learning_rate": 3.638765882307589e-05,
      "loss": 10.271,
      "step": 1082
    },
    {
      "epoch": 0.11868737028843059,
      "grad_norm": 0.0507303848862648,
      "learning_rate": 3.6225114898308634e-05,
      "loss": 10.2736,
      "step": 1083
    },
    {
      "epoch": 0.11879696158140236,
      "grad_norm": 0.047596875578165054,
      "learning_rate": 3.6062854488992714e-05,
      "loss": 10.2793,
      "step": 1084
    },
    {
      "epoch": 0.11890655287437413,
      "grad_norm": 0.0487726628780365,
      "learning_rate": 3.5900878316467454e-05,
      "loss": 10.2704,
      "step": 1085
    },
    {
      "epoch": 0.1190161441673459,
      "grad_norm": 0.051695920526981354,
      "learning_rate": 3.573918710080857e-05,
      "loss": 10.275,
      "step": 1086
    },
    {
      "epoch": 0.11912573546031768,
      "grad_norm": 0.06416413187980652,
      "learning_rate": 3.5577781560825066e-05,
      "loss": 10.2718,
      "step": 1087
    },
    {
      "epoch": 0.11923532675328945,
      "grad_norm": 0.048725325614213943,
      "learning_rate": 3.541666241405588e-05,
      "loss": 10.2675,
      "step": 1088
    },
    {
      "epoch": 0.11934491804626123,
      "grad_norm": 0.05156528949737549,
      "learning_rate": 3.5255830376766764e-05,
      "loss": 10.2677,
      "step": 1089
    },
    {
      "epoch": 0.119454509339233,
      "grad_norm": 0.04582783579826355,
      "learning_rate": 3.509528616394716e-05,
      "loss": 10.2726,
      "step": 1090
    },
    {
      "epoch": 0.11956410063220477,
      "grad_norm": 0.048340681940317154,
      "learning_rate": 3.4935030489306883e-05,
      "loss": 10.2675,
      "step": 1091
    },
    {
      "epoch": 0.11967369192517655,
      "grad_norm": 0.06420094519853592,
      "learning_rate": 3.4775064065273165e-05,
      "loss": 10.2623,
      "step": 1092
    },
    {
      "epoch": 0.11978328321814832,
      "grad_norm": 0.04663741961121559,
      "learning_rate": 3.4615387602987236e-05,
      "loss": 10.2717,
      "step": 1093
    },
    {
      "epoch": 0.11989287451112009,
      "grad_norm": 0.05812327563762665,
      "learning_rate": 3.445600181230134e-05,
      "loss": 10.2786,
      "step": 1094
    },
    {
      "epoch": 0.12000246580409187,
      "grad_norm": 0.056189801543951035,
      "learning_rate": 3.429690740177549e-05,
      "loss": 10.2715,
      "step": 1095
    },
    {
      "epoch": 0.12011205709706364,
      "grad_norm": 0.07483986020088196,
      "learning_rate": 3.413810507867436e-05,
      "loss": 10.2738,
      "step": 1096
    },
    {
      "epoch": 0.12022164839003541,
      "grad_norm": 0.05344222113490105,
      "learning_rate": 3.397959554896415e-05,
      "loss": 10.2721,
      "step": 1097
    },
    {
      "epoch": 0.12033123968300719,
      "grad_norm": 0.049546632915735245,
      "learning_rate": 3.3821379517309405e-05,
      "loss": 10.2676,
      "step": 1098
    },
    {
      "epoch": 0.12044083097597896,
      "grad_norm": 0.07524579018354416,
      "learning_rate": 3.3663457687069924e-05,
      "loss": 10.2759,
      "step": 1099
    },
    {
      "epoch": 0.12055042226895073,
      "grad_norm": 0.0568055734038353,
      "learning_rate": 3.350583076029754e-05,
      "loss": 10.2758,
      "step": 1100
    },
    {
      "epoch": 0.1206600135619225,
      "grad_norm": 0.03264664113521576,
      "learning_rate": 3.334849943773323e-05,
      "loss": 10.2768,
      "step": 1101
    },
    {
      "epoch": 0.12076960485489428,
      "grad_norm": 0.03622061014175415,
      "learning_rate": 3.319146441880371e-05,
      "loss": 10.2766,
      "step": 1102
    },
    {
      "epoch": 0.12087919614786605,
      "grad_norm": 0.03506995737552643,
      "learning_rate": 3.3034726401618444e-05,
      "loss": 10.2763,
      "step": 1103
    },
    {
      "epoch": 0.12098878744083782,
      "grad_norm": 0.04481271654367447,
      "learning_rate": 3.28782860829667e-05,
      "loss": 10.2807,
      "step": 1104
    },
    {
      "epoch": 0.1210983787338096,
      "grad_norm": 0.046395059674978256,
      "learning_rate": 3.272214415831418e-05,
      "loss": 10.2731,
      "step": 1105
    },
    {
      "epoch": 0.12120797002678137,
      "grad_norm": 0.047015026211738586,
      "learning_rate": 3.2566301321800085e-05,
      "loss": 10.2796,
      "step": 1106
    },
    {
      "epoch": 0.12131756131975314,
      "grad_norm": 0.03600795194506645,
      "learning_rate": 3.241075826623401e-05,
      "loss": 10.2768,
      "step": 1107
    },
    {
      "epoch": 0.12142715261272492,
      "grad_norm": 0.04693186655640602,
      "learning_rate": 3.225551568309284e-05,
      "loss": 10.2824,
      "step": 1108
    },
    {
      "epoch": 0.12153674390569669,
      "grad_norm": 0.0421341247856617,
      "learning_rate": 3.210057426251773e-05,
      "loss": 10.2764,
      "step": 1109
    },
    {
      "epoch": 0.12164633519866846,
      "grad_norm": 0.04485126957297325,
      "learning_rate": 3.1945934693310896e-05,
      "loss": 10.2818,
      "step": 1110
    },
    {
      "epoch": 0.12175592649164024,
      "grad_norm": 0.043029844760894775,
      "learning_rate": 3.179159766293282e-05,
      "loss": 10.2757,
      "step": 1111
    },
    {
      "epoch": 0.121865517784612,
      "grad_norm": 0.04897027835249901,
      "learning_rate": 3.163756385749889e-05,
      "loss": 10.2727,
      "step": 1112
    },
    {
      "epoch": 0.12197510907758378,
      "grad_norm": 0.05395561829209328,
      "learning_rate": 3.148383396177653e-05,
      "loss": 10.286,
      "step": 1113
    },
    {
      "epoch": 0.12208470037055556,
      "grad_norm": 0.04522831365466118,
      "learning_rate": 3.133040865918213e-05,
      "loss": 10.2784,
      "step": 1114
    },
    {
      "epoch": 0.12219429166352733,
      "grad_norm": 0.04812748730182648,
      "learning_rate": 3.117728863177796e-05,
      "loss": 10.2664,
      "step": 1115
    },
    {
      "epoch": 0.1223038829564991,
      "grad_norm": 0.04504070803523064,
      "learning_rate": 3.102447456026919e-05,
      "loss": 10.2851,
      "step": 1116
    },
    {
      "epoch": 0.12241347424947088,
      "grad_norm": 0.0470891110599041,
      "learning_rate": 3.0871967124000834e-05,
      "loss": 10.2778,
      "step": 1117
    },
    {
      "epoch": 0.12252306554244265,
      "grad_norm": 0.04827373847365379,
      "learning_rate": 3.0719767000954714e-05,
      "loss": 10.2711,
      "step": 1118
    },
    {
      "epoch": 0.12263265683541442,
      "grad_norm": 0.04363931342959404,
      "learning_rate": 3.056787486774656e-05,
      "loss": 10.2764,
      "step": 1119
    },
    {
      "epoch": 0.1227422481283862,
      "grad_norm": 0.03992973640561104,
      "learning_rate": 3.041629139962283e-05,
      "loss": 10.2725,
      "step": 1120
    },
    {
      "epoch": 0.12285183942135797,
      "grad_norm": 0.04178993031382561,
      "learning_rate": 3.0265017270457775e-05,
      "loss": 10.2754,
      "step": 1121
    },
    {
      "epoch": 0.12296143071432974,
      "grad_norm": 0.046450383961200714,
      "learning_rate": 3.0114053152750556e-05,
      "loss": 10.272,
      "step": 1122
    },
    {
      "epoch": 0.12307102200730152,
      "grad_norm": 0.04929455742239952,
      "learning_rate": 2.9963399717622077e-05,
      "loss": 10.276,
      "step": 1123
    },
    {
      "epoch": 0.12318061330027329,
      "grad_norm": 0.048688508570194244,
      "learning_rate": 2.98130576348121e-05,
      "loss": 10.2802,
      "step": 1124
    },
    {
      "epoch": 0.12329020459324506,
      "grad_norm": 0.04904251545667648,
      "learning_rate": 2.966302757267625e-05,
      "loss": 10.2795,
      "step": 1125
    },
    {
      "epoch": 0.12329020459324506,
      "eval_loss": 10.27295970916748,
      "eval_runtime": 53.582,
      "eval_samples_per_second": 286.831,
      "eval_steps_per_second": 143.425,
      "step": 1125
    },
    {
      "epoch": 0.12339979588621684,
      "grad_norm": 0.05245248228311539,
      "learning_rate": 2.9513310198183065e-05,
      "loss": 10.2583,
      "step": 1126
    },
    {
      "epoch": 0.1235093871791886,
      "grad_norm": 0.0569288544356823,
      "learning_rate": 2.936390617691097e-05,
      "loss": 10.2702,
      "step": 1127
    },
    {
      "epoch": 0.12361897847216038,
      "grad_norm": 0.04590731859207153,
      "learning_rate": 2.9214816173045356e-05,
      "loss": 10.2722,
      "step": 1128
    },
    {
      "epoch": 0.12372856976513216,
      "grad_norm": 0.04473479464650154,
      "learning_rate": 2.906604084937572e-05,
      "loss": 10.2693,
      "step": 1129
    },
    {
      "epoch": 0.12383816105810394,
      "grad_norm": 0.05548057332634926,
      "learning_rate": 2.8917580867292526e-05,
      "loss": 10.2672,
      "step": 1130
    },
    {
      "epoch": 0.1239477523510757,
      "grad_norm": 0.05406226962804794,
      "learning_rate": 2.8769436886784408e-05,
      "loss": 10.2616,
      "step": 1131
    },
    {
      "epoch": 0.12405734364404748,
      "grad_norm": 0.04907669126987457,
      "learning_rate": 2.862160956643517e-05,
      "loss": 10.269,
      "step": 1132
    },
    {
      "epoch": 0.12416693493701926,
      "grad_norm": 0.04661229997873306,
      "learning_rate": 2.847409956342092e-05,
      "loss": 10.2713,
      "step": 1133
    },
    {
      "epoch": 0.12427652622999102,
      "grad_norm": 0.048107411712408066,
      "learning_rate": 2.8326907533507074e-05,
      "loss": 10.2753,
      "step": 1134
    },
    {
      "epoch": 0.1243861175229628,
      "grad_norm": 0.052953027188777924,
      "learning_rate": 2.8180034131045464e-05,
      "loss": 10.2649,
      "step": 1135
    },
    {
      "epoch": 0.12449570881593458,
      "grad_norm": 0.05090215057134628,
      "learning_rate": 2.8033480008971546e-05,
      "loss": 10.2753,
      "step": 1136
    },
    {
      "epoch": 0.12460530010890634,
      "grad_norm": 0.054354727268218994,
      "learning_rate": 2.7887245818801277e-05,
      "loss": 10.2628,
      "step": 1137
    },
    {
      "epoch": 0.12471489140187812,
      "grad_norm": 0.05736605077981949,
      "learning_rate": 2.7741332210628345e-05,
      "loss": 10.2783,
      "step": 1138
    },
    {
      "epoch": 0.1248244826948499,
      "grad_norm": 0.048060912638902664,
      "learning_rate": 2.759573983312138e-05,
      "loss": 10.2736,
      "step": 1139
    },
    {
      "epoch": 0.12493407398782166,
      "grad_norm": 0.059280525892972946,
      "learning_rate": 2.7450469333520855e-05,
      "loss": 10.2679,
      "step": 1140
    },
    {
      "epoch": 0.12504366528079344,
      "grad_norm": 0.04672105237841606,
      "learning_rate": 2.730552135763632e-05,
      "loss": 10.2755,
      "step": 1141
    },
    {
      "epoch": 0.12515325657376522,
      "grad_norm": 0.05562569573521614,
      "learning_rate": 2.7160896549843562e-05,
      "loss": 10.2754,
      "step": 1142
    },
    {
      "epoch": 0.125262847866737,
      "grad_norm": 0.05804011970758438,
      "learning_rate": 2.701659555308169e-05,
      "loss": 10.2686,
      "step": 1143
    },
    {
      "epoch": 0.12537243915970875,
      "grad_norm": 0.05067833140492439,
      "learning_rate": 2.6872619008850274e-05,
      "loss": 10.2705,
      "step": 1144
    },
    {
      "epoch": 0.12548203045268053,
      "grad_norm": 0.05379986763000488,
      "learning_rate": 2.672896755720654e-05,
      "loss": 10.2594,
      "step": 1145
    },
    {
      "epoch": 0.1255916217456523,
      "grad_norm": 0.05436115339398384,
      "learning_rate": 2.6585641836762433e-05,
      "loss": 10.2579,
      "step": 1146
    },
    {
      "epoch": 0.12570121303862408,
      "grad_norm": 0.05774844065308571,
      "learning_rate": 2.6442642484681944e-05,
      "loss": 10.2712,
      "step": 1147
    },
    {
      "epoch": 0.12581080433159586,
      "grad_norm": 0.0658312439918518,
      "learning_rate": 2.6299970136678077e-05,
      "loss": 10.2659,
      "step": 1148
    },
    {
      "epoch": 0.12592039562456764,
      "grad_norm": 0.06053013727068901,
      "learning_rate": 2.6157625427010156e-05,
      "loss": 10.2765,
      "step": 1149
    },
    {
      "epoch": 0.1260299869175394,
      "grad_norm": 0.06674255430698395,
      "learning_rate": 2.6015608988480955e-05,
      "loss": 10.2757,
      "step": 1150
    },
    {
      "epoch": 0.12613957821051117,
      "grad_norm": 0.03407241404056549,
      "learning_rate": 2.5873921452433915e-05,
      "loss": 10.2812,
      "step": 1151
    },
    {
      "epoch": 0.12624916950348294,
      "grad_norm": 0.04747309163212776,
      "learning_rate": 2.57325634487503e-05,
      "loss": 10.2826,
      "step": 1152
    },
    {
      "epoch": 0.12635876079645472,
      "grad_norm": 0.04173409193754196,
      "learning_rate": 2.5591535605846383e-05,
      "loss": 10.284,
      "step": 1153
    },
    {
      "epoch": 0.1264683520894265,
      "grad_norm": 0.04084724187850952,
      "learning_rate": 2.5450838550670808e-05,
      "loss": 10.282,
      "step": 1154
    },
    {
      "epoch": 0.12657794338239828,
      "grad_norm": 0.05204812064766884,
      "learning_rate": 2.5310472908701555e-05,
      "loss": 10.2837,
      "step": 1155
    },
    {
      "epoch": 0.12668753467537003,
      "grad_norm": 0.05498720705509186,
      "learning_rate": 2.5170439303943294e-05,
      "loss": 10.2802,
      "step": 1156
    },
    {
      "epoch": 0.1267971259683418,
      "grad_norm": 0.03891552612185478,
      "learning_rate": 2.503073835892471e-05,
      "loss": 10.2684,
      "step": 1157
    },
    {
      "epoch": 0.12690671726131358,
      "grad_norm": 0.04726177826523781,
      "learning_rate": 2.4891370694695517e-05,
      "loss": 10.2751,
      "step": 1158
    },
    {
      "epoch": 0.12701630855428536,
      "grad_norm": 0.04519693925976753,
      "learning_rate": 2.4752336930823837e-05,
      "loss": 10.263,
      "step": 1159
    },
    {
      "epoch": 0.12712589984725714,
      "grad_norm": 0.057225536555051804,
      "learning_rate": 2.4613637685393432e-05,
      "loss": 10.2743,
      "step": 1160
    },
    {
      "epoch": 0.12723549114022892,
      "grad_norm": 0.06272728741168976,
      "learning_rate": 2.4475273575000936e-05,
      "loss": 10.2715,
      "step": 1161
    },
    {
      "epoch": 0.1273450824332007,
      "grad_norm": 0.04253806918859482,
      "learning_rate": 2.4337245214753103e-05,
      "loss": 10.279,
      "step": 1162
    },
    {
      "epoch": 0.12745467372617245,
      "grad_norm": 0.055703457444906235,
      "learning_rate": 2.4199553218264093e-05,
      "loss": 10.2779,
      "step": 1163
    },
    {
      "epoch": 0.12756426501914422,
      "grad_norm": 0.06134609133005142,
      "learning_rate": 2.4062198197652752e-05,
      "loss": 10.2783,
      "step": 1164
    },
    {
      "epoch": 0.127673856312116,
      "grad_norm": 0.057871509343385696,
      "learning_rate": 2.3925180763539844e-05,
      "loss": 10.2809,
      "step": 1165
    },
    {
      "epoch": 0.12778344760508778,
      "grad_norm": 0.044840008020401,
      "learning_rate": 2.3788501525045438e-05,
      "loss": 10.279,
      "step": 1166
    },
    {
      "epoch": 0.12789303889805956,
      "grad_norm": 0.036749910563230515,
      "learning_rate": 2.3652161089786086e-05,
      "loss": 10.2702,
      "step": 1167
    },
    {
      "epoch": 0.12800263019103134,
      "grad_norm": 0.05096394941210747,
      "learning_rate": 2.351616006387214e-05,
      "loss": 10.2878,
      "step": 1168
    },
    {
      "epoch": 0.12811222148400309,
      "grad_norm": 0.050375353544950485,
      "learning_rate": 2.3380499051905137e-05,
      "loss": 10.2806,
      "step": 1169
    },
    {
      "epoch": 0.12822181277697486,
      "grad_norm": 0.05090630054473877,
      "learning_rate": 2.324517865697501e-05,
      "loss": 10.2683,
      "step": 1170
    },
    {
      "epoch": 0.12833140406994664,
      "grad_norm": 0.04255339503288269,
      "learning_rate": 2.3110199480657525e-05,
      "loss": 10.266,
      "step": 1171
    },
    {
      "epoch": 0.12844099536291842,
      "grad_norm": 0.05159114673733711,
      "learning_rate": 2.2975562123011495e-05,
      "loss": 10.2708,
      "step": 1172
    },
    {
      "epoch": 0.1285505866558902,
      "grad_norm": 0.03814525157213211,
      "learning_rate": 2.2841267182576143e-05,
      "loss": 10.2826,
      "step": 1173
    },
    {
      "epoch": 0.12866017794886198,
      "grad_norm": 0.046735979616642,
      "learning_rate": 2.2707315256368433e-05,
      "loss": 10.2698,
      "step": 1174
    },
    {
      "epoch": 0.12876976924183373,
      "grad_norm": 0.04669572412967682,
      "learning_rate": 2.2573706939880555e-05,
      "loss": 10.2624,
      "step": 1175
    },
    {
      "epoch": 0.1288793605348055,
      "grad_norm": 0.047829918563365936,
      "learning_rate": 2.2440442827077045e-05,
      "loss": 10.2821,
      "step": 1176
    },
    {
      "epoch": 0.12898895182777728,
      "grad_norm": 0.04206380993127823,
      "learning_rate": 2.230752351039228e-05,
      "loss": 10.2766,
      "step": 1177
    },
    {
      "epoch": 0.12909854312074906,
      "grad_norm": 0.042676832526922226,
      "learning_rate": 2.2174949580727832e-05,
      "loss": 10.2757,
      "step": 1178
    },
    {
      "epoch": 0.12920813441372084,
      "grad_norm": 0.054205987602472305,
      "learning_rate": 2.2042721627449846e-05,
      "loss": 10.2801,
      "step": 1179
    },
    {
      "epoch": 0.12931772570669262,
      "grad_norm": 0.06192367151379585,
      "learning_rate": 2.1910840238386398e-05,
      "loss": 10.2694,
      "step": 1180
    },
    {
      "epoch": 0.12942731699966437,
      "grad_norm": 0.046615634113550186,
      "learning_rate": 2.1779305999824884e-05,
      "loss": 10.2779,
      "step": 1181
    },
    {
      "epoch": 0.12953690829263614,
      "grad_norm": 0.0508783683180809,
      "learning_rate": 2.164811949650942e-05,
      "loss": 10.2707,
      "step": 1182
    },
    {
      "epoch": 0.12964649958560792,
      "grad_norm": 0.048583000898361206,
      "learning_rate": 2.1517281311638217e-05,
      "loss": 10.2831,
      "step": 1183
    },
    {
      "epoch": 0.1297560908785797,
      "grad_norm": 0.05060219019651413,
      "learning_rate": 2.1386792026861103e-05,
      "loss": 10.2702,
      "step": 1184
    },
    {
      "epoch": 0.12986568217155148,
      "grad_norm": 0.056559156626462936,
      "learning_rate": 2.125665222227675e-05,
      "loss": 10.2764,
      "step": 1185
    },
    {
      "epoch": 0.12997527346452326,
      "grad_norm": 0.05017093941569328,
      "learning_rate": 2.112686247643024e-05,
      "loss": 10.2592,
      "step": 1186
    },
    {
      "epoch": 0.130084864757495,
      "grad_norm": 0.054072678089141846,
      "learning_rate": 2.09974233663104e-05,
      "loss": 10.2688,
      "step": 1187
    },
    {
      "epoch": 0.13019445605046678,
      "grad_norm": 0.050235409289598465,
      "learning_rate": 2.0868335467347366e-05,
      "loss": 10.2709,
      "step": 1188
    },
    {
      "epoch": 0.13030404734343856,
      "grad_norm": 0.053272001445293427,
      "learning_rate": 2.073959935340988e-05,
      "loss": 10.2673,
      "step": 1189
    },
    {
      "epoch": 0.13041363863641034,
      "grad_norm": 0.04364420846104622,
      "learning_rate": 2.06112155968028e-05,
      "loss": 10.2755,
      "step": 1190
    },
    {
      "epoch": 0.13052322992938212,
      "grad_norm": 0.05550595000386238,
      "learning_rate": 2.0483184768264596e-05,
      "loss": 10.2721,
      "step": 1191
    },
    {
      "epoch": 0.1306328212223539,
      "grad_norm": 0.047221601009368896,
      "learning_rate": 2.035550743696468e-05,
      "loss": 10.2751,
      "step": 1192
    },
    {
      "epoch": 0.13074241251532565,
      "grad_norm": 0.0528549924492836,
      "learning_rate": 2.022818417050113e-05,
      "loss": 10.2729,
      "step": 1193
    },
    {
      "epoch": 0.13085200380829742,
      "grad_norm": 0.05821693688631058,
      "learning_rate": 2.0101215534897855e-05,
      "loss": 10.2741,
      "step": 1194
    },
    {
      "epoch": 0.1309615951012692,
      "grad_norm": 0.05353042855858803,
      "learning_rate": 1.99746020946023e-05,
      "loss": 10.2686,
      "step": 1195
    },
    {
      "epoch": 0.13107118639424098,
      "grad_norm": 0.052468787878751755,
      "learning_rate": 1.9848344412482854e-05,
      "loss": 10.2699,
      "step": 1196
    },
    {
      "epoch": 0.13118077768721276,
      "grad_norm": 0.06008310243487358,
      "learning_rate": 1.9722443049826344e-05,
      "loss": 10.2733,
      "step": 1197
    },
    {
      "epoch": 0.13129036898018454,
      "grad_norm": 0.06696496158838272,
      "learning_rate": 1.9596898566335576e-05,
      "loss": 10.269,
      "step": 1198
    },
    {
      "epoch": 0.13139996027315629,
      "grad_norm": 0.04962735250592232,
      "learning_rate": 1.9471711520126824e-05,
      "loss": 10.2798,
      "step": 1199
    },
    {
      "epoch": 0.13150955156612806,
      "grad_norm": 0.06983551383018494,
      "learning_rate": 1.9346882467727325e-05,
      "loss": 10.2824,
      "step": 1200
    },
    {
      "epoch": 0.13161914285909984,
      "grad_norm": 0.04193832352757454,
      "learning_rate": 1.9222411964072884e-05,
      "loss": 10.2749,
      "step": 1201
    },
    {
      "epoch": 0.13172873415207162,
      "grad_norm": 0.03930622339248657,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 10.2766,
      "step": 1202
    },
    {
      "epoch": 0.1318383254450434,
      "grad_norm": 0.04357234016060829,
      "learning_rate": 1.8974548814769944e-05,
      "loss": 10.2732,
      "step": 1203
    },
    {
      "epoch": 0.13194791673801518,
      "grad_norm": 0.04687696695327759,
      "learning_rate": 1.8851157271013442e-05,
      "loss": 10.2784,
      "step": 1204
    },
    {
      "epoch": 0.13205750803098693,
      "grad_norm": 0.045309461653232574,
      "learning_rate": 1.872812647978095e-05,
      "loss": 10.2807,
      "step": 1205
    },
    {
      "epoch": 0.1321670993239587,
      "grad_norm": 0.051966022700071335,
      "learning_rate": 1.8605456988014015e-05,
      "loss": 10.2767,
      "step": 1206
    },
    {
      "epoch": 0.13227669061693048,
      "grad_norm": 0.04756461828947067,
      "learning_rate": 1.8483149341047923e-05,
      "loss": 10.2833,
      "step": 1207
    },
    {
      "epoch": 0.13238628190990226,
      "grad_norm": 0.044760771095752716,
      "learning_rate": 1.8361204082609352e-05,
      "loss": 10.2785,
      "step": 1208
    },
    {
      "epoch": 0.13249587320287404,
      "grad_norm": 0.046154871582984924,
      "learning_rate": 1.8239621754813995e-05,
      "loss": 10.27,
      "step": 1209
    },
    {
      "epoch": 0.13260546449584582,
      "grad_norm": 0.05204872414469719,
      "learning_rate": 1.811840289816409e-05,
      "loss": 10.2706,
      "step": 1210
    },
    {
      "epoch": 0.13271505578881757,
      "grad_norm": 0.0465998537838459,
      "learning_rate": 1.799754805154603e-05,
      "loss": 10.27,
      "step": 1211
    },
    {
      "epoch": 0.13282464708178934,
      "grad_norm": 0.05250348523259163,
      "learning_rate": 1.787705775222802e-05,
      "loss": 10.2797,
      "step": 1212
    },
    {
      "epoch": 0.13293423837476112,
      "grad_norm": 0.060549307614564896,
      "learning_rate": 1.775693253585763e-05,
      "loss": 10.2716,
      "step": 1213
    },
    {
      "epoch": 0.1330438296677329,
      "grad_norm": 0.0557357482612133,
      "learning_rate": 1.763717293645939e-05,
      "loss": 10.2725,
      "step": 1214
    },
    {
      "epoch": 0.13315342096070468,
      "grad_norm": 0.0737464427947998,
      "learning_rate": 1.7517779486432495e-05,
      "loss": 10.2869,
      "step": 1215
    },
    {
      "epoch": 0.13326301225367645,
      "grad_norm": 0.06550686806440353,
      "learning_rate": 1.7398752716548395e-05,
      "loss": 10.2754,
      "step": 1216
    },
    {
      "epoch": 0.1333726035466482,
      "grad_norm": 0.04716264456510544,
      "learning_rate": 1.728009315594843e-05,
      "loss": 10.2797,
      "step": 1217
    },
    {
      "epoch": 0.13348219483961998,
      "grad_norm": 0.05071970447897911,
      "learning_rate": 1.716180133214149e-05,
      "loss": 10.2766,
      "step": 1218
    },
    {
      "epoch": 0.13359178613259176,
      "grad_norm": 0.04534721001982689,
      "learning_rate": 1.704387777100165e-05,
      "loss": 10.2713,
      "step": 1219
    },
    {
      "epoch": 0.13370137742556354,
      "grad_norm": 0.04618574678897858,
      "learning_rate": 1.6926322996765897e-05,
      "loss": 10.2764,
      "step": 1220
    },
    {
      "epoch": 0.13381096871853532,
      "grad_norm": 0.05128798633813858,
      "learning_rate": 1.6809137532031704e-05,
      "loss": 10.2824,
      "step": 1221
    },
    {
      "epoch": 0.1339205600115071,
      "grad_norm": 0.053042635321617126,
      "learning_rate": 1.6692321897754758e-05,
      "loss": 10.2631,
      "step": 1222
    },
    {
      "epoch": 0.13403015130447887,
      "grad_norm": 0.04228586331009865,
      "learning_rate": 1.65758766132467e-05,
      "loss": 10.2803,
      "step": 1223
    },
    {
      "epoch": 0.13413974259745062,
      "grad_norm": 0.06437396258115768,
      "learning_rate": 1.6459802196172668e-05,
      "loss": 10.2691,
      "step": 1224
    },
    {
      "epoch": 0.1342493338904224,
      "grad_norm": 0.04839962720870972,
      "learning_rate": 1.634409916254914e-05,
      "loss": 10.2737,
      "step": 1225
    },
    {
      "epoch": 0.13435892518339418,
      "grad_norm": 0.04871037229895592,
      "learning_rate": 1.622876802674158e-05,
      "loss": 10.2739,
      "step": 1226
    },
    {
      "epoch": 0.13446851647636596,
      "grad_norm": 0.044558003544807434,
      "learning_rate": 1.6113809301462125e-05,
      "loss": 10.2727,
      "step": 1227
    },
    {
      "epoch": 0.13457810776933773,
      "grad_norm": 0.04532990604639053,
      "learning_rate": 1.599922349776738e-05,
      "loss": 10.2684,
      "step": 1228
    },
    {
      "epoch": 0.1346876990623095,
      "grad_norm": 0.041408050805330276,
      "learning_rate": 1.5885011125056047e-05,
      "loss": 10.2644,
      "step": 1229
    },
    {
      "epoch": 0.13479729035528126,
      "grad_norm": 0.050078023225069046,
      "learning_rate": 1.5771172691066794e-05,
      "loss": 10.2695,
      "step": 1230
    },
    {
      "epoch": 0.13490688164825304,
      "grad_norm": 0.04859043285250664,
      "learning_rate": 1.565770870187585e-05,
      "loss": 10.2744,
      "step": 1231
    },
    {
      "epoch": 0.13501647294122482,
      "grad_norm": 0.05008580908179283,
      "learning_rate": 1.5544619661894864e-05,
      "loss": 10.2689,
      "step": 1232
    },
    {
      "epoch": 0.1351260642341966,
      "grad_norm": 0.04766001179814339,
      "learning_rate": 1.543190607386861e-05,
      "loss": 10.2766,
      "step": 1233
    },
    {
      "epoch": 0.13523565552716837,
      "grad_norm": 0.04672755300998688,
      "learning_rate": 1.5319568438872745e-05,
      "loss": 10.2831,
      "step": 1234
    },
    {
      "epoch": 0.13534524682014015,
      "grad_norm": 0.05103520303964615,
      "learning_rate": 1.520760725631164e-05,
      "loss": 10.273,
      "step": 1235
    },
    {
      "epoch": 0.1354548381131119,
      "grad_norm": 0.05670294538140297,
      "learning_rate": 1.5096023023916094e-05,
      "loss": 10.2632,
      "step": 1236
    },
    {
      "epoch": 0.13556442940608368,
      "grad_norm": 0.048262566328048706,
      "learning_rate": 1.498481623774115e-05,
      "loss": 10.2641,
      "step": 1237
    },
    {
      "epoch": 0.13567402069905546,
      "grad_norm": 0.06024409458041191,
      "learning_rate": 1.4873987392163947e-05,
      "loss": 10.2737,
      "step": 1238
    },
    {
      "epoch": 0.13578361199202724,
      "grad_norm": 0.05486240237951279,
      "learning_rate": 1.4763536979881354e-05,
      "loss": 10.2791,
      "step": 1239
    },
    {
      "epoch": 0.13589320328499901,
      "grad_norm": 0.06305771321058273,
      "learning_rate": 1.4653465491908003e-05,
      "loss": 10.2638,
      "step": 1240
    },
    {
      "epoch": 0.1360027945779708,
      "grad_norm": 0.04569699242711067,
      "learning_rate": 1.4543773417573925e-05,
      "loss": 10.268,
      "step": 1241
    },
    {
      "epoch": 0.13611238587094254,
      "grad_norm": 0.040898025035858154,
      "learning_rate": 1.4434461244522458e-05,
      "loss": 10.2687,
      "step": 1242
    },
    {
      "epoch": 0.13622197716391432,
      "grad_norm": 0.05176094174385071,
      "learning_rate": 1.4325529458708065e-05,
      "loss": 10.2747,
      "step": 1243
    },
    {
      "epoch": 0.1363315684568861,
      "grad_norm": 0.05542001128196716,
      "learning_rate": 1.4216978544394177e-05,
      "loss": 10.2738,
      "step": 1244
    },
    {
      "epoch": 0.13644115974985788,
      "grad_norm": 0.058724407106637955,
      "learning_rate": 1.4108808984151023e-05,
      "loss": 10.2709,
      "step": 1245
    },
    {
      "epoch": 0.13655075104282965,
      "grad_norm": 0.04667770862579346,
      "learning_rate": 1.4001021258853509e-05,
      "loss": 10.2656,
      "step": 1246
    },
    {
      "epoch": 0.13666034233580143,
      "grad_norm": 0.062016263604164124,
      "learning_rate": 1.3893615847679065e-05,
      "loss": 10.2756,
      "step": 1247
    },
    {
      "epoch": 0.13676993362877318,
      "grad_norm": 0.05500452220439911,
      "learning_rate": 1.3786593228105494e-05,
      "loss": 10.2799,
      "step": 1248
    },
    {
      "epoch": 0.13687952492174496,
      "grad_norm": 0.05758122354745865,
      "learning_rate": 1.3679953875908957e-05,
      "loss": 10.274,
      "step": 1249
    },
    {
      "epoch": 0.13698911621471674,
      "grad_norm": 0.06983663886785507,
      "learning_rate": 1.3573698265161683e-05,
      "loss": 10.2816,
      "step": 1250
    },
    {
      "epoch": 0.13709870750768852,
      "grad_norm": 0.03310950845479965,
      "learning_rate": 1.3467826868229994e-05,
      "loss": 10.2849,
      "step": 1251
    },
    {
      "epoch": 0.1372082988006603,
      "grad_norm": 0.034070901572704315,
      "learning_rate": 1.3362340155772146e-05,
      "loss": 10.2735,
      "step": 1252
    },
    {
      "epoch": 0.13731789009363207,
      "grad_norm": 0.048867177218198776,
      "learning_rate": 1.3257238596736266e-05,
      "loss": 10.2795,
      "step": 1253
    },
    {
      "epoch": 0.13742748138660382,
      "grad_norm": 0.04260621592402458,
      "learning_rate": 1.3152522658358245e-05,
      "loss": 10.2843,
      "step": 1254
    },
    {
      "epoch": 0.1375370726795756,
      "grad_norm": 0.04184035584330559,
      "learning_rate": 1.3048192806159721e-05,
      "loss": 10.2766,
      "step": 1255
    },
    {
      "epoch": 0.13764666397254738,
      "grad_norm": 0.052505455911159515,
      "learning_rate": 1.2944249503945894e-05,
      "loss": 10.2799,
      "step": 1256
    },
    {
      "epoch": 0.13775625526551916,
      "grad_norm": 0.0542524978518486,
      "learning_rate": 1.2840693213803545e-05,
      "loss": 10.2757,
      "step": 1257
    },
    {
      "epoch": 0.13786584655849093,
      "grad_norm": 0.04224352911114693,
      "learning_rate": 1.2737524396099032e-05,
      "loss": 10.2745,
      "step": 1258
    },
    {
      "epoch": 0.1379754378514627,
      "grad_norm": 0.06141772121191025,
      "learning_rate": 1.2634743509476088e-05,
      "loss": 10.28,
      "step": 1259
    },
    {
      "epoch": 0.13808502914443446,
      "grad_norm": 0.05849756672978401,
      "learning_rate": 1.2532351010853916e-05,
      "loss": 10.2666,
      "step": 1260
    },
    {
      "epoch": 0.13819462043740624,
      "grad_norm": 0.06311602145433426,
      "learning_rate": 1.243034735542512e-05,
      "loss": 10.2803,
      "step": 1261
    },
    {
      "epoch": 0.13830421173037802,
      "grad_norm": 0.06097150221467018,
      "learning_rate": 1.2328732996653669e-05,
      "loss": 10.2747,
      "step": 1262
    },
    {
      "epoch": 0.1384138030233498,
      "grad_norm": 0.059486374258995056,
      "learning_rate": 1.2227508386272878e-05,
      "loss": 10.2777,
      "step": 1263
    },
    {
      "epoch": 0.13852339431632157,
      "grad_norm": 0.056624118238687515,
      "learning_rate": 1.212667397428342e-05,
      "loss": 10.2766,
      "step": 1264
    },
    {
      "epoch": 0.13863298560929335,
      "grad_norm": 0.055356208235025406,
      "learning_rate": 1.2026230208951306e-05,
      "loss": 10.2802,
      "step": 1265
    },
    {
      "epoch": 0.1387425769022651,
      "grad_norm": 0.05898386612534523,
      "learning_rate": 1.1926177536805905e-05,
      "loss": 10.2846,
      "step": 1266
    },
    {
      "epoch": 0.13885216819523688,
      "grad_norm": 0.05229266360402107,
      "learning_rate": 1.1826516402637989e-05,
      "loss": 10.2749,
      "step": 1267
    },
    {
      "epoch": 0.13896175948820866,
      "grad_norm": 0.0455329604446888,
      "learning_rate": 1.1727247249497685e-05,
      "loss": 10.271,
      "step": 1268
    },
    {
      "epoch": 0.13907135078118044,
      "grad_norm": 0.04478120058774948,
      "learning_rate": 1.1628370518692533e-05,
      "loss": 10.2709,
      "step": 1269
    },
    {
      "epoch": 0.13918094207415221,
      "grad_norm": 0.04799001291394234,
      "learning_rate": 1.152988664978556e-05,
      "loss": 10.2715,
      "step": 1270
    },
    {
      "epoch": 0.139290533367124,
      "grad_norm": 0.04392573609948158,
      "learning_rate": 1.1431796080593283e-05,
      "loss": 10.2674,
      "step": 1271
    },
    {
      "epoch": 0.13940012466009574,
      "grad_norm": 0.05076228454709053,
      "learning_rate": 1.1334099247183783e-05,
      "loss": 10.2688,
      "step": 1272
    },
    {
      "epoch": 0.13950971595306752,
      "grad_norm": 0.04884777218103409,
      "learning_rate": 1.1236796583874787e-05,
      "loss": 10.2695,
      "step": 1273
    },
    {
      "epoch": 0.1396193072460393,
      "grad_norm": 0.049333445727825165,
      "learning_rate": 1.1139888523231678e-05,
      "loss": 10.2693,
      "step": 1274
    },
    {
      "epoch": 0.13972889853901108,
      "grad_norm": 0.04502782225608826,
      "learning_rate": 1.1043375496065611e-05,
      "loss": 10.274,
      "step": 1275
    },
    {
      "epoch": 0.13983848983198285,
      "grad_norm": 0.042851049453020096,
      "learning_rate": 1.0947257931431642e-05,
      "loss": 10.2753,
      "step": 1276
    },
    {
      "epoch": 0.13994808112495463,
      "grad_norm": 0.04958132281899452,
      "learning_rate": 1.0851536256626705e-05,
      "loss": 10.272,
      "step": 1277
    },
    {
      "epoch": 0.14005767241792638,
      "grad_norm": 0.0428130067884922,
      "learning_rate": 1.0756210897187812e-05,
      "loss": 10.271,
      "step": 1278
    },
    {
      "epoch": 0.14016726371089816,
      "grad_norm": 0.04856995493173599,
      "learning_rate": 1.0661282276890127e-05,
      "loss": 10.269,
      "step": 1279
    },
    {
      "epoch": 0.14027685500386994,
      "grad_norm": 0.0523076169192791,
      "learning_rate": 1.0566750817745074e-05,
      "loss": 10.2778,
      "step": 1280
    },
    {
      "epoch": 0.14038644629684172,
      "grad_norm": 0.042247019708156586,
      "learning_rate": 1.0472616939998492e-05,
      "loss": 10.2809,
      "step": 1281
    },
    {
      "epoch": 0.1404960375898135,
      "grad_norm": 0.04855649173259735,
      "learning_rate": 1.0378881062128731e-05,
      "loss": 10.2649,
      "step": 1282
    },
    {
      "epoch": 0.14060562888278527,
      "grad_norm": 0.05721621587872505,
      "learning_rate": 1.0285543600844804e-05,
      "loss": 10.2669,
      "step": 1283
    },
    {
      "epoch": 0.14071522017575705,
      "grad_norm": 0.05200272426009178,
      "learning_rate": 1.019260497108453e-05,
      "loss": 10.2767,
      "step": 1284
    },
    {
      "epoch": 0.1408248114687288,
      "grad_norm": 0.03965900465846062,
      "learning_rate": 1.010006558601274e-05,
      "loss": 10.2657,
      "step": 1285
    },
    {
      "epoch": 0.14093440276170058,
      "grad_norm": 0.047540392726659775,
      "learning_rate": 1.000792585701934e-05,
      "loss": 10.2694,
      "step": 1286
    },
    {
      "epoch": 0.14104399405467236,
      "grad_norm": 0.04648537188768387,
      "learning_rate": 9.91618619371757e-06,
      "loss": 10.2695,
      "step": 1287
    },
    {
      "epoch": 0.14115358534764413,
      "grad_norm": 0.05335364118218422,
      "learning_rate": 9.82484700394215e-06,
      "loss": 10.2724,
      "step": 1288
    },
    {
      "epoch": 0.1412631766406159,
      "grad_norm": 0.056220006197690964,
      "learning_rate": 9.73390869374743e-06,
      "loss": 10.2719,
      "step": 1289
    },
    {
      "epoch": 0.1413727679335877,
      "grad_norm": 0.04979753866791725,
      "learning_rate": 9.643371667405698e-06,
      "loss": 10.2694,
      "step": 1290
    },
    {
      "epoch": 0.14148235922655944,
      "grad_norm": 0.050429701805114746,
      "learning_rate": 9.553236327405246e-06,
      "loss": 10.2617,
      "step": 1291
    },
    {
      "epoch": 0.14159195051953122,
      "grad_norm": 0.049937453120946884,
      "learning_rate": 9.463503074448677e-06,
      "loss": 10.271,
      "step": 1292
    },
    {
      "epoch": 0.141701541812503,
      "grad_norm": 0.05065898597240448,
      "learning_rate": 9.374172307451068e-06,
      "loss": 10.2672,
      "step": 1293
    },
    {
      "epoch": 0.14181113310547477,
      "grad_norm": 0.04540201276540756,
      "learning_rate": 9.285244423538197e-06,
      "loss": 10.2654,
      "step": 1294
    },
    {
      "epoch": 0.14192072439844655,
      "grad_norm": 0.06275458633899689,
      "learning_rate": 9.196719818044886e-06,
      "loss": 10.2667,
      "step": 1295
    },
    {
      "epoch": 0.14203031569141833,
      "grad_norm": 0.04951043426990509,
      "learning_rate": 9.108598884513053e-06,
      "loss": 10.2611,
      "step": 1296
    },
    {
      "epoch": 0.14213990698439008,
      "grad_norm": 0.05271495506167412,
      "learning_rate": 9.020882014690136e-06,
      "loss": 10.2744,
      "step": 1297
    },
    {
      "epoch": 0.14224949827736186,
      "grad_norm": 0.059048768132925034,
      "learning_rate": 8.933569598527247e-06,
      "loss": 10.2667,
      "step": 1298
    },
    {
      "epoch": 0.14235908957033364,
      "grad_norm": 0.0689985454082489,
      "learning_rate": 8.846662024177477e-06,
      "loss": 10.2782,
      "step": 1299
    },
    {
      "epoch": 0.1424686808633054,
      "grad_norm": 0.06539405882358551,
      "learning_rate": 8.760159677994172e-06,
      "loss": 10.274,
      "step": 1300
    },
    {
      "epoch": 0.1425782721562772,
      "grad_norm": 0.0367017462849617,
      "learning_rate": 8.674062944529216e-06,
      "loss": 10.2751,
      "step": 1301
    },
    {
      "epoch": 0.14268786344924897,
      "grad_norm": 0.03982223570346832,
      "learning_rate": 8.588372206531292e-06,
      "loss": 10.2777,
      "step": 1302
    },
    {
      "epoch": 0.14279745474222072,
      "grad_norm": 0.037616245448589325,
      "learning_rate": 8.503087844944213e-06,
      "loss": 10.2779,
      "step": 1303
    },
    {
      "epoch": 0.1429070460351925,
      "grad_norm": 0.0494188629090786,
      "learning_rate": 8.418210238905256e-06,
      "loss": 10.2715,
      "step": 1304
    },
    {
      "epoch": 0.14301663732816428,
      "grad_norm": 0.04668893665075302,
      "learning_rate": 8.333739765743398e-06,
      "loss": 10.2729,
      "step": 1305
    },
    {
      "epoch": 0.14312622862113605,
      "grad_norm": 0.04076192528009415,
      "learning_rate": 8.249676800977658e-06,
      "loss": 10.2766,
      "step": 1306
    },
    {
      "epoch": 0.14323581991410783,
      "grad_norm": 0.0486617274582386,
      "learning_rate": 8.16602171831553e-06,
      "loss": 10.28,
      "step": 1307
    },
    {
      "epoch": 0.1433454112070796,
      "grad_norm": 0.04288191720843315,
      "learning_rate": 8.082774889651168e-06,
      "loss": 10.2741,
      "step": 1308
    },
    {
      "epoch": 0.14345500250005136,
      "grad_norm": 0.045139130204916,
      "learning_rate": 7.999936685063835e-06,
      "loss": 10.2758,
      "step": 1309
    },
    {
      "epoch": 0.14356459379302314,
      "grad_norm": 0.044365499168634415,
      "learning_rate": 7.91750747281621e-06,
      "loss": 10.2739,
      "step": 1310
    },
    {
      "epoch": 0.14367418508599492,
      "grad_norm": 0.05269256979227066,
      "learning_rate": 7.835487619352811e-06,
      "loss": 10.2755,
      "step": 1311
    },
    {
      "epoch": 0.1437837763789667,
      "grad_norm": 0.04736129567027092,
      "learning_rate": 7.753877489298244e-06,
      "loss": 10.282,
      "step": 1312
    },
    {
      "epoch": 0.14389336767193847,
      "grad_norm": 0.05673832818865776,
      "learning_rate": 7.67267744545579e-06,
      "loss": 10.2808,
      "step": 1313
    },
    {
      "epoch": 0.14400295896491025,
      "grad_norm": 0.06118704378604889,
      "learning_rate": 7.591887848805545e-06,
      "loss": 10.2813,
      "step": 1314
    },
    {
      "epoch": 0.144112550257882,
      "grad_norm": 0.06224576011300087,
      "learning_rate": 7.5115090585029966e-06,
      "loss": 10.2882,
      "step": 1315
    },
    {
      "epoch": 0.14422214155085378,
      "grad_norm": 0.06372867524623871,
      "learning_rate": 7.431541431877342e-06,
      "loss": 10.2777,
      "step": 1316
    },
    {
      "epoch": 0.14433173284382556,
      "grad_norm": 0.04780254140496254,
      "learning_rate": 7.351985324429933e-06,
      "loss": 10.2713,
      "step": 1317
    },
    {
      "epoch": 0.14444132413679733,
      "grad_norm": 0.046562857925891876,
      "learning_rate": 7.272841089832694e-06,
      "loss": 10.2749,
      "step": 1318
    },
    {
      "epoch": 0.1445509154297691,
      "grad_norm": 0.05586632341146469,
      "learning_rate": 7.194109079926514e-06,
      "loss": 10.2812,
      "step": 1319
    },
    {
      "epoch": 0.1446605067227409,
      "grad_norm": 0.05409142002463341,
      "learning_rate": 7.115789644719728e-06,
      "loss": 10.2849,
      "step": 1320
    },
    {
      "epoch": 0.14477009801571264,
      "grad_norm": 0.05445246398448944,
      "learning_rate": 7.037883132386547e-06,
      "loss": 10.2644,
      "step": 1321
    },
    {
      "epoch": 0.14487968930868442,
      "grad_norm": 0.046961914747953415,
      "learning_rate": 6.960389889265517e-06,
      "loss": 10.2665,
      "step": 1322
    },
    {
      "epoch": 0.1449892806016562,
      "grad_norm": 0.060011934489011765,
      "learning_rate": 6.883310259857944e-06,
      "loss": 10.2669,
      "step": 1323
    },
    {
      "epoch": 0.14509887189462797,
      "grad_norm": 0.053605277091264725,
      "learning_rate": 6.806644586826383e-06,
      "loss": 10.2682,
      "step": 1324
    },
    {
      "epoch": 0.14520846318759975,
      "grad_norm": 0.050913531333208084,
      "learning_rate": 6.730393210993147e-06,
      "loss": 10.2707,
      "step": 1325
    },
    {
      "epoch": 0.14531805448057153,
      "grad_norm": 0.050232261419296265,
      "learning_rate": 6.654556471338746e-06,
      "loss": 10.2695,
      "step": 1326
    },
    {
      "epoch": 0.14542764577354328,
      "grad_norm": 0.05975714325904846,
      "learning_rate": 6.579134705000412e-06,
      "loss": 10.2704,
      "step": 1327
    },
    {
      "epoch": 0.14553723706651506,
      "grad_norm": 0.04994051530957222,
      "learning_rate": 6.504128247270546e-06,
      "loss": 10.272,
      "step": 1328
    },
    {
      "epoch": 0.14564682835948684,
      "grad_norm": 0.04960297420620918,
      "learning_rate": 6.429537431595312e-06,
      "loss": 10.2767,
      "step": 1329
    },
    {
      "epoch": 0.1457564196524586,
      "grad_norm": 0.048075269907712936,
      "learning_rate": 6.355362589573077e-06,
      "loss": 10.2656,
      "step": 1330
    },
    {
      "epoch": 0.1458660109454304,
      "grad_norm": 0.061504118144512177,
      "learning_rate": 6.2816040509530165e-06,
      "loss": 10.2735,
      "step": 1331
    },
    {
      "epoch": 0.14597560223840217,
      "grad_norm": 0.05464784428477287,
      "learning_rate": 6.2082621436335475e-06,
      "loss": 10.2692,
      "step": 1332
    },
    {
      "epoch": 0.14608519353137392,
      "grad_norm": 0.052297502756118774,
      "learning_rate": 6.135337193660962e-06,
      "loss": 10.264,
      "step": 1333
    },
    {
      "epoch": 0.1461947848243457,
      "grad_norm": 0.04761944338679314,
      "learning_rate": 6.062829525227909e-06,
      "loss": 10.2666,
      "step": 1334
    },
    {
      "epoch": 0.14630437611731748,
      "grad_norm": 0.04858838766813278,
      "learning_rate": 5.990739460672024e-06,
      "loss": 10.2642,
      "step": 1335
    },
    {
      "epoch": 0.14641396741028925,
      "grad_norm": 0.04581503942608833,
      "learning_rate": 5.9190673204744255e-06,
      "loss": 10.2776,
      "step": 1336
    },
    {
      "epoch": 0.14652355870326103,
      "grad_norm": 0.048014722764492035,
      "learning_rate": 5.84781342325833e-06,
      "loss": 10.2644,
      "step": 1337
    },
    {
      "epoch": 0.1466331499962328,
      "grad_norm": 0.054411496967077255,
      "learning_rate": 5.77697808578761e-06,
      "loss": 10.2741,
      "step": 1338
    },
    {
      "epoch": 0.14674274128920456,
      "grad_norm": 0.058360274881124496,
      "learning_rate": 5.706561622965467e-06,
      "loss": 10.2689,
      "step": 1339
    },
    {
      "epoch": 0.14685233258217634,
      "grad_norm": 0.047847211360931396,
      "learning_rate": 5.636564347832907e-06,
      "loss": 10.2698,
      "step": 1340
    },
    {
      "epoch": 0.14696192387514811,
      "grad_norm": 0.05588516592979431,
      "learning_rate": 5.566986571567401e-06,
      "loss": 10.2666,
      "step": 1341
    },
    {
      "epoch": 0.1470715151681199,
      "grad_norm": 0.06412149965763092,
      "learning_rate": 5.497828603481569e-06,
      "loss": 10.2663,
      "step": 1342
    },
    {
      "epoch": 0.14718110646109167,
      "grad_norm": 0.05627089738845825,
      "learning_rate": 5.429090751021704e-06,
      "loss": 10.2687,
      "step": 1343
    },
    {
      "epoch": 0.14729069775406345,
      "grad_norm": 0.05770175904035568,
      "learning_rate": 5.3607733197664436e-06,
      "loss": 10.2742,
      "step": 1344
    },
    {
      "epoch": 0.14740028904703523,
      "grad_norm": 0.05818668380379677,
      "learning_rate": 5.2928766134254345e-06,
      "loss": 10.272,
      "step": 1345
    },
    {
      "epoch": 0.14750988034000698,
      "grad_norm": 0.05893673375248909,
      "learning_rate": 5.225400933837954e-06,
      "loss": 10.269,
      "step": 1346
    },
    {
      "epoch": 0.14761947163297875,
      "grad_norm": 0.06425610184669495,
      "learning_rate": 5.158346580971573e-06,
      "loss": 10.278,
      "step": 1347
    },
    {
      "epoch": 0.14772906292595053,
      "grad_norm": 0.059625208377838135,
      "learning_rate": 5.091713852920854e-06,
      "loss": 10.2629,
      "step": 1348
    },
    {
      "epoch": 0.1478386542189223,
      "grad_norm": 0.06345664709806442,
      "learning_rate": 5.025503045905933e-06,
      "loss": 10.2779,
      "step": 1349
    },
    {
      "epoch": 0.1479482455118941,
      "grad_norm": 0.0641244575381279,
      "learning_rate": 4.959714454271369e-06,
      "loss": 10.2796,
      "step": 1350
    },
    {
      "epoch": 0.14805783680486587,
      "grad_norm": 0.033460378646850586,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 10.2746,
      "step": 1351
    },
    {
      "epoch": 0.14816742809783762,
      "grad_norm": 0.054631225764751434,
      "learning_rate": 4.829405085134997e-06,
      "loss": 10.2823,
      "step": 1352
    },
    {
      "epoch": 0.1482770193908094,
      "grad_norm": 0.037300437688827515,
      "learning_rate": 4.764884886932086e-06,
      "loss": 10.2855,
      "step": 1353
    },
    {
      "epoch": 0.14838661068378117,
      "grad_norm": 0.04847235232591629,
      "learning_rate": 4.700788062704687e-06,
      "loss": 10.2813,
      "step": 1354
    },
    {
      "epoch": 0.14849620197675295,
      "grad_norm": 0.041923802345991135,
      "learning_rate": 4.6371148973994525e-06,
      "loss": 10.2738,
      "step": 1355
    },
    {
      "epoch": 0.14860579326972473,
      "grad_norm": 0.04788297042250633,
      "learning_rate": 4.573865674079625e-06,
      "loss": 10.2836,
      "step": 1356
    },
    {
      "epoch": 0.1487153845626965,
      "grad_norm": 0.045485515147447586,
      "learning_rate": 4.511040673923828e-06,
      "loss": 10.2769,
      "step": 1357
    },
    {
      "epoch": 0.14882497585566826,
      "grad_norm": 0.04665224626660347,
      "learning_rate": 4.448640176224694e-06,
      "loss": 10.2752,
      "step": 1358
    },
    {
      "epoch": 0.14893456714864003,
      "grad_norm": 0.05267065390944481,
      "learning_rate": 4.386664458387779e-06,
      "loss": 10.2727,
      "step": 1359
    },
    {
      "epoch": 0.1490441584416118,
      "grad_norm": 0.04310646653175354,
      "learning_rate": 4.325113795930203e-06,
      "loss": 10.272,
      "step": 1360
    },
    {
      "epoch": 0.1491537497345836,
      "grad_norm": 0.051386523991823196,
      "learning_rate": 4.263988462479484e-06,
      "loss": 10.2716,
      "step": 1361
    },
    {
      "epoch": 0.14926334102755537,
      "grad_norm": 0.04798987880349159,
      "learning_rate": 4.203288729772326e-06,
      "loss": 10.2692,
      "step": 1362
    },
    {
      "epoch": 0.14937293232052715,
      "grad_norm": 0.06597252190113068,
      "learning_rate": 4.143014867653383e-06,
      "loss": 10.2763,
      "step": 1363
    },
    {
      "epoch": 0.1494825236134989,
      "grad_norm": 0.06321340799331665,
      "learning_rate": 4.083167144074073e-06,
      "loss": 10.2714,
      "step": 1364
    },
    {
      "epoch": 0.14959211490647067,
      "grad_norm": 0.07340846210718155,
      "learning_rate": 4.023745825091407e-06,
      "loss": 10.287,
      "step": 1365
    },
    {
      "epoch": 0.14970170619944245,
      "grad_norm": 0.05240330845117569,
      "learning_rate": 3.964751174866765e-06,
      "loss": 10.2782,
      "step": 1366
    },
    {
      "epoch": 0.14981129749241423,
      "grad_norm": 0.04362007975578308,
      "learning_rate": 3.906183455664725e-06,
      "loss": 10.266,
      "step": 1367
    },
    {
      "epoch": 0.149920888785386,
      "grad_norm": 0.04652499035000801,
      "learning_rate": 3.84804292785198e-06,
      "loss": 10.2803,
      "step": 1368
    },
    {
      "epoch": 0.1500304800783578,
      "grad_norm": 0.040402717888355255,
      "learning_rate": 3.7903298498960572e-06,
      "loss": 10.2701,
      "step": 1369
    },
    {
      "epoch": 0.15014007137132954,
      "grad_norm": 0.04453877732157707,
      "learning_rate": 3.7330444783642338e-06,
      "loss": 10.2703,
      "step": 1370
    },
    {
      "epoch": 0.15024966266430131,
      "grad_norm": 0.056760355830192566,
      "learning_rate": 3.676187067922421e-06,
      "loss": 10.2804,
      "step": 1371
    },
    {
      "epoch": 0.1503592539572731,
      "grad_norm": 0.05622982606291771,
      "learning_rate": 3.619757871333973e-06,
      "loss": 10.2755,
      "step": 1372
    },
    {
      "epoch": 0.15046884525024487,
      "grad_norm": 0.050346702337265015,
      "learning_rate": 3.563757139458579e-06,
      "loss": 10.2764,
      "step": 1373
    },
    {
      "epoch": 0.15057843654321665,
      "grad_norm": 0.03686307743191719,
      "learning_rate": 3.5081851212512175e-06,
      "loss": 10.2778,
      "step": 1374
    },
    {
      "epoch": 0.15068802783618843,
      "grad_norm": 0.050080060958862305,
      "learning_rate": 3.4530420637609363e-06,
      "loss": 10.2673,
      "step": 1375
    },
    {
      "epoch": 0.15079761912916018,
      "grad_norm": 0.04181602597236633,
      "learning_rate": 3.3983282121298086e-06,
      "loss": 10.2722,
      "step": 1376
    },
    {
      "epoch": 0.15090721042213195,
      "grad_norm": 0.03709720820188522,
      "learning_rate": 3.3440438095919126e-06,
      "loss": 10.277,
      "step": 1377
    },
    {
      "epoch": 0.15101680171510373,
      "grad_norm": 0.06096412241458893,
      "learning_rate": 3.290189097472096e-06,
      "loss": 10.2585,
      "step": 1378
    },
    {
      "epoch": 0.1511263930080755,
      "grad_norm": 0.043854936957359314,
      "learning_rate": 3.236764315185037e-06,
      "loss": 10.2747,
      "step": 1379
    },
    {
      "epoch": 0.1512359843010473,
      "grad_norm": 0.0491960272192955,
      "learning_rate": 3.1837697002341293e-06,
      "loss": 10.2714,
      "step": 1380
    },
    {
      "epoch": 0.15134557559401907,
      "grad_norm": 0.049867551773786545,
      "learning_rate": 3.131205488210409e-06,
      "loss": 10.2712,
      "step": 1381
    },
    {
      "epoch": 0.15145516688699082,
      "grad_norm": 0.053189218044281006,
      "learning_rate": 3.0790719127915646e-06,
      "loss": 10.2696,
      "step": 1382
    },
    {
      "epoch": 0.1515647581799626,
      "grad_norm": 0.05457506328821182,
      "learning_rate": 3.0273692057408265e-06,
      "loss": 10.2618,
      "step": 1383
    },
    {
      "epoch": 0.15167434947293437,
      "grad_norm": 0.046817876398563385,
      "learning_rate": 2.976097596905969e-06,
      "loss": 10.2682,
      "step": 1384
    },
    {
      "epoch": 0.15178394076590615,
      "grad_norm": 0.058962829411029816,
      "learning_rate": 2.9252573142183326e-06,
      "loss": 10.2815,
      "step": 1385
    },
    {
      "epoch": 0.15189353205887793,
      "grad_norm": 0.04770894721150398,
      "learning_rate": 2.874848583691714e-06,
      "loss": 10.2682,
      "step": 1386
    },
    {
      "epoch": 0.1520031233518497,
      "grad_norm": 0.053199172019958496,
      "learning_rate": 2.8248716294214774e-06,
      "loss": 10.2663,
      "step": 1387
    },
    {
      "epoch": 0.15211271464482146,
      "grad_norm": 0.04962414503097534,
      "learning_rate": 2.7753266735834338e-06,
      "loss": 10.2686,
      "step": 1388
    },
    {
      "epoch": 0.15222230593779323,
      "grad_norm": 0.0513363815844059,
      "learning_rate": 2.7262139364329643e-06,
      "loss": 10.2599,
      "step": 1389
    },
    {
      "epoch": 0.152331897230765,
      "grad_norm": 0.04942251741886139,
      "learning_rate": 2.677533636303964e-06,
      "loss": 10.2815,
      "step": 1390
    },
    {
      "epoch": 0.1524414885237368,
      "grad_norm": 0.04966640844941139,
      "learning_rate": 2.6292859896079213e-06,
      "loss": 10.2738,
      "step": 1391
    },
    {
      "epoch": 0.15255107981670857,
      "grad_norm": 0.05927044153213501,
      "learning_rate": 2.581471210832931e-06,
      "loss": 10.2646,
      "step": 1392
    },
    {
      "epoch": 0.15266067110968035,
      "grad_norm": 0.055439066141843796,
      "learning_rate": 2.5340895125427364e-06,
      "loss": 10.2666,
      "step": 1393
    },
    {
      "epoch": 0.1527702624026521,
      "grad_norm": 0.054749272763729095,
      "learning_rate": 2.4871411053757898e-06,
      "loss": 10.2728,
      "step": 1394
    },
    {
      "epoch": 0.15287985369562387,
      "grad_norm": 0.06369620561599731,
      "learning_rate": 2.440626198044327e-06,
      "loss": 10.2622,
      "step": 1395
    },
    {
      "epoch": 0.15298944498859565,
      "grad_norm": 0.059688955545425415,
      "learning_rate": 2.394544997333437e-06,
      "loss": 10.271,
      "step": 1396
    },
    {
      "epoch": 0.15309903628156743,
      "grad_norm": 0.05762460082769394,
      "learning_rate": 2.3488977081001394e-06,
      "loss": 10.281,
      "step": 1397
    },
    {
      "epoch": 0.1532086275745392,
      "grad_norm": 0.060228921473026276,
      "learning_rate": 2.3036845332724543e-06,
      "loss": 10.2702,
      "step": 1398
    },
    {
      "epoch": 0.15331821886751099,
      "grad_norm": 0.07223130762577057,
      "learning_rate": 2.2589056738485324e-06,
      "loss": 10.2719,
      "step": 1399
    },
    {
      "epoch": 0.15342781016048274,
      "grad_norm": 0.06956610083580017,
      "learning_rate": 2.2145613288957478e-06,
      "loss": 10.2741,
      "step": 1400
    },
    {
      "epoch": 0.15353740145345451,
      "grad_norm": 0.03410380706191063,
      "learning_rate": 2.170651695549786e-06,
      "loss": 10.2803,
      "step": 1401
    },
    {
      "epoch": 0.1536469927464263,
      "grad_norm": 0.033173032104969025,
      "learning_rate": 2.1271769690138332e-06,
      "loss": 10.274,
      "step": 1402
    },
    {
      "epoch": 0.15375658403939807,
      "grad_norm": 0.04483955726027489,
      "learning_rate": 2.084137342557646e-06,
      "loss": 10.2769,
      "step": 1403
    },
    {
      "epoch": 0.15386617533236985,
      "grad_norm": 0.0385802797973156,
      "learning_rate": 2.0415330075166937e-06,
      "loss": 10.2768,
      "step": 1404
    },
    {
      "epoch": 0.15397576662534163,
      "grad_norm": 0.041432302445173264,
      "learning_rate": 1.9993641532913833e-06,
      "loss": 10.2761,
      "step": 1405
    },
    {
      "epoch": 0.1540853579183134,
      "grad_norm": 0.05089664086699486,
      "learning_rate": 1.9576309673461357e-06,
      "loss": 10.2791,
      "step": 1406
    },
    {
      "epoch": 0.15419494921128515,
      "grad_norm": 0.03721743822097778,
      "learning_rate": 1.916333635208556e-06,
      "loss": 10.271,
      "step": 1407
    },
    {
      "epoch": 0.15430454050425693,
      "grad_norm": 0.054213616997003555,
      "learning_rate": 1.8754723404686425e-06,
      "loss": 10.2761,
      "step": 1408
    },
    {
      "epoch": 0.1544141317972287,
      "grad_norm": 0.05848401412367821,
      "learning_rate": 1.8350472647780116e-06,
      "loss": 10.2819,
      "step": 1409
    },
    {
      "epoch": 0.1545237230902005,
      "grad_norm": 0.048858508467674255,
      "learning_rate": 1.7950585878489856e-06,
      "loss": 10.2738,
      "step": 1410
    },
    {
      "epoch": 0.15463331438317227,
      "grad_norm": 0.05895935744047165,
      "learning_rate": 1.7555064874538397e-06,
      "loss": 10.2771,
      "step": 1411
    },
    {
      "epoch": 0.15474290567614404,
      "grad_norm": 0.05079033970832825,
      "learning_rate": 1.7163911394240672e-06,
      "loss": 10.2807,
      "step": 1412
    },
    {
      "epoch": 0.1548524969691158,
      "grad_norm": 0.06598055362701416,
      "learning_rate": 1.6777127176495043e-06,
      "loss": 10.2797,
      "step": 1413
    },
    {
      "epoch": 0.15496208826208757,
      "grad_norm": 0.053515248000621796,
      "learning_rate": 1.6394713940776296e-06,
      "loss": 10.285,
      "step": 1414
    },
    {
      "epoch": 0.15507167955505935,
      "grad_norm": 0.06508594751358032,
      "learning_rate": 1.6016673387127646e-06,
      "loss": 10.2809,
      "step": 1415
    },
    {
      "epoch": 0.15518127084803113,
      "grad_norm": 0.04955340921878815,
      "learning_rate": 1.5643007196153302e-06,
      "loss": 10.2696,
      "step": 1416
    },
    {
      "epoch": 0.1552908621410029,
      "grad_norm": 0.04417774826288223,
      "learning_rate": 1.5273717029010925e-06,
      "loss": 10.2752,
      "step": 1417
    },
    {
      "epoch": 0.15540045343397468,
      "grad_norm": 0.04010520502924919,
      "learning_rate": 1.4908804527404286e-06,
      "loss": 10.2806,
      "step": 1418
    },
    {
      "epoch": 0.15551004472694643,
      "grad_norm": 0.04590655863285065,
      "learning_rate": 1.4548271313575835e-06,
      "loss": 10.2764,
      "step": 1419
    },
    {
      "epoch": 0.1556196360199182,
      "grad_norm": 0.05384558439254761,
      "learning_rate": 1.4192118990299707e-06,
      "loss": 10.2741,
      "step": 1420
    },
    {
      "epoch": 0.15572922731289,
      "grad_norm": 0.04424852132797241,
      "learning_rate": 1.3840349140874619e-06,
      "loss": 10.268,
      "step": 1421
    },
    {
      "epoch": 0.15583881860586177,
      "grad_norm": 0.04343269020318985,
      "learning_rate": 1.3492963329116537e-06,
      "loss": 10.2725,
      "step": 1422
    },
    {
      "epoch": 0.15594840989883355,
      "grad_norm": 0.05392204597592354,
      "learning_rate": 1.3149963099352014e-06,
      "loss": 10.2764,
      "step": 1423
    },
    {
      "epoch": 0.15605800119180532,
      "grad_norm": 0.047185055911540985,
      "learning_rate": 1.2811349976411202e-06,
      "loss": 10.2681,
      "step": 1424
    },
    {
      "epoch": 0.15616759248477707,
      "grad_norm": 0.04843048006296158,
      "learning_rate": 1.2477125465620853e-06,
      "loss": 10.2675,
      "step": 1425
    },
    {
      "epoch": 0.15627718377774885,
      "grad_norm": 0.0545368418097496,
      "learning_rate": 1.2147291052798216e-06,
      "loss": 10.2741,
      "step": 1426
    },
    {
      "epoch": 0.15638677507072063,
      "grad_norm": 0.039745207875967026,
      "learning_rate": 1.1821848204243814e-06,
      "loss": 10.2717,
      "step": 1427
    },
    {
      "epoch": 0.1564963663636924,
      "grad_norm": 0.0489082932472229,
      "learning_rate": 1.1500798366735233e-06,
      "loss": 10.2668,
      "step": 1428
    },
    {
      "epoch": 0.15660595765666419,
      "grad_norm": 0.049913935363292694,
      "learning_rate": 1.1184142967520794e-06,
      "loss": 10.2786,
      "step": 1429
    },
    {
      "epoch": 0.15671554894963596,
      "grad_norm": 0.03998143598437309,
      "learning_rate": 1.0871883414312777e-06,
      "loss": 10.2692,
      "step": 1430
    },
    {
      "epoch": 0.1568251402426077,
      "grad_norm": 0.06323674321174622,
      "learning_rate": 1.0564021095281652e-06,
      "loss": 10.2713,
      "step": 1431
    },
    {
      "epoch": 0.1569347315355795,
      "grad_norm": 0.04850061610341072,
      "learning_rate": 1.0260557379049519e-06,
      "loss": 10.2622,
      "step": 1432
    },
    {
      "epoch": 0.15704432282855127,
      "grad_norm": 0.06200779229402542,
      "learning_rate": 9.96149361468457e-07,
      "loss": 10.2731,
      "step": 1433
    },
    {
      "epoch": 0.15715391412152305,
      "grad_norm": 0.052363522350788116,
      "learning_rate": 9.66683113169431e-07,
      "loss": 10.2747,
      "step": 1434
    },
    {
      "epoch": 0.15726350541449483,
      "grad_norm": 0.05521099641919136,
      "learning_rate": 9.376571240020227e-07,
      "loss": 10.2645,
      "step": 1435
    },
    {
      "epoch": 0.1573730967074666,
      "grad_norm": 0.04292723163962364,
      "learning_rate": 9.090715230031688e-07,
      "loss": 10.2612,
      "step": 1436
    },
    {
      "epoch": 0.15748268800043835,
      "grad_norm": 0.05433180555701256,
      "learning_rate": 8.809264372520609e-07,
      "loss": 10.2604,
      "step": 1437
    },
    {
      "epoch": 0.15759227929341013,
      "grad_norm": 0.04529083892703056,
      "learning_rate": 8.532219918695128e-07,
      "loss": 10.2833,
      "step": 1438
    },
    {
      "epoch": 0.1577018705863819,
      "grad_norm": 0.04527038708329201,
      "learning_rate": 8.259583100174606e-07,
      "loss": 10.2639,
      "step": 1439
    },
    {
      "epoch": 0.1578114618793537,
      "grad_norm": 0.0496191531419754,
      "learning_rate": 7.991355128984079e-07,
      "loss": 10.273,
      "step": 1440
    },
    {
      "epoch": 0.15792105317232547,
      "grad_norm": 0.05632885918021202,
      "learning_rate": 7.727537197548707e-07,
      "loss": 10.2714,
      "step": 1441
    },
    {
      "epoch": 0.15803064446529724,
      "grad_norm": 0.058481018990278244,
      "learning_rate": 7.468130478688218e-07,
      "loss": 10.267,
      "step": 1442
    },
    {
      "epoch": 0.158140235758269,
      "grad_norm": 0.04811238497495651,
      "learning_rate": 7.213136125612586e-07,
      "loss": 10.2758,
      "step": 1443
    },
    {
      "epoch": 0.15824982705124077,
      "grad_norm": 0.048781946301460266,
      "learning_rate": 6.962555271915805e-07,
      "loss": 10.2551,
      "step": 1444
    },
    {
      "epoch": 0.15835941834421255,
      "grad_norm": 0.065910205245018,
      "learning_rate": 6.716389031571568e-07,
      "loss": 10.2755,
      "step": 1445
    },
    {
      "epoch": 0.15846900963718433,
      "grad_norm": 0.052832599729299545,
      "learning_rate": 6.474638498928265e-07,
      "loss": 10.267,
      "step": 1446
    },
    {
      "epoch": 0.1585786009301561,
      "grad_norm": 0.047852519899606705,
      "learning_rate": 6.237304748703543e-07,
      "loss": 10.2706,
      "step": 1447
    },
    {
      "epoch": 0.15868819222312788,
      "grad_norm": 0.059627044945955276,
      "learning_rate": 6.004388835980423e-07,
      "loss": 10.2801,
      "step": 1448
    },
    {
      "epoch": 0.15879778351609963,
      "grad_norm": 0.05848207324743271,
      "learning_rate": 5.77589179620186e-07,
      "loss": 10.2709,
      "step": 1449
    },
    {
      "epoch": 0.1589073748090714,
      "grad_norm": 0.06132341921329498,
      "learning_rate": 5.55181464516652e-07,
      "loss": 10.2838,
      "step": 1450
    },
    {
      "epoch": 0.1590169661020432,
      "grad_norm": 0.034376852214336395,
      "learning_rate": 5.332158379024122e-07,
      "loss": 10.2784,
      "step": 1451
    },
    {
      "epoch": 0.15912655739501497,
      "grad_norm": 0.03615598753094673,
      "learning_rate": 5.116923974270993e-07,
      "loss": 10.2772,
      "step": 1452
    },
    {
      "epoch": 0.15923614868798674,
      "grad_norm": 0.03944362699985504,
      "learning_rate": 4.906112387745965e-07,
      "loss": 10.2811,
      "step": 1453
    },
    {
      "epoch": 0.15934573998095852,
      "grad_norm": 0.043196823447942734,
      "learning_rate": 4.6997245566257064e-07,
      "loss": 10.2793,
      "step": 1454
    },
    {
      "epoch": 0.15945533127393027,
      "grad_norm": 0.04201686382293701,
      "learning_rate": 4.497761398421063e-07,
      "loss": 10.2742,
      "step": 1455
    },
    {
      "epoch": 0.15956492256690205,
      "grad_norm": 0.04621388763189316,
      "learning_rate": 4.3002238109723927e-07,
      "loss": 10.2717,
      "step": 1456
    },
    {
      "epoch": 0.15967451385987383,
      "grad_norm": 0.046749334782361984,
      "learning_rate": 4.107112672446123e-07,
      "loss": 10.2783,
      "step": 1457
    },
    {
      "epoch": 0.1597841051528456,
      "grad_norm": 0.04872756078839302,
      "learning_rate": 3.9184288413306456e-07,
      "loss": 10.279,
      "step": 1458
    },
    {
      "epoch": 0.15989369644581738,
      "grad_norm": 0.05072105675935745,
      "learning_rate": 3.734173156432208e-07,
      "loss": 10.2851,
      "step": 1459
    },
    {
      "epoch": 0.16000328773878916,
      "grad_norm": 0.050555698573589325,
      "learning_rate": 3.554346436871581e-07,
      "loss": 10.2729,
      "step": 1460
    },
    {
      "epoch": 0.1601128790317609,
      "grad_norm": 0.04541072994470596,
      "learning_rate": 3.3789494820803957e-07,
      "loss": 10.2692,
      "step": 1461
    },
    {
      "epoch": 0.1602224703247327,
      "grad_norm": 0.0657145082950592,
      "learning_rate": 3.2079830717972606e-07,
      "loss": 10.2732,
      "step": 1462
    },
    {
      "epoch": 0.16033206161770447,
      "grad_norm": 0.060773175209760666,
      "learning_rate": 3.041447966064648e-07,
      "loss": 10.2787,
      "step": 1463
    },
    {
      "epoch": 0.16044165291067625,
      "grad_norm": 0.07818783074617386,
      "learning_rate": 2.8793449052254563e-07,
      "loss": 10.2791,
      "step": 1464
    },
    {
      "epoch": 0.16055124420364802,
      "grad_norm": 0.07057569175958633,
      "learning_rate": 2.721674609919345e-07,
      "loss": 10.2832,
      "step": 1465
    },
    {
      "epoch": 0.1606608354966198,
      "grad_norm": 0.045935992151498795,
      "learning_rate": 2.568437781080069e-07,
      "loss": 10.2739,
      "step": 1466
    },
    {
      "epoch": 0.16077042678959158,
      "grad_norm": 0.052126262336969376,
      "learning_rate": 2.4196350999320384e-07,
      "loss": 10.2739,
      "step": 1467
    },
    {
      "epoch": 0.16088001808256333,
      "grad_norm": 0.04580167308449745,
      "learning_rate": 2.275267227987321e-07,
      "loss": 10.264,
      "step": 1468
    },
    {
      "epoch": 0.1609896093755351,
      "grad_norm": 0.053641047328710556,
      "learning_rate": 2.135334807042866e-07,
      "loss": 10.2843,
      "step": 1469
    },
    {
      "epoch": 0.1610992006685069,
      "grad_norm": 0.050128720700740814,
      "learning_rate": 1.9998384591773944e-07,
      "loss": 10.2728,
      "step": 1470
    },
    {
      "epoch": 0.16120879196147866,
      "grad_norm": 0.05269044265151024,
      "learning_rate": 1.8687787867489592e-07,
      "loss": 10.2765,
      "step": 1471
    },
    {
      "epoch": 0.16131838325445044,
      "grad_norm": 0.057393208146095276,
      "learning_rate": 1.7421563723919454e-07,
      "loss": 10.2714,
      "step": 1472
    },
    {
      "epoch": 0.16142797454742222,
      "grad_norm": 0.051147978752851486,
      "learning_rate": 1.6199717790145174e-07,
      "loss": 10.2711,
      "step": 1473
    },
    {
      "epoch": 0.16153756584039397,
      "grad_norm": 0.06040102243423462,
      "learning_rate": 1.5022255497962879e-07,
      "loss": 10.2759,
      "step": 1474
    },
    {
      "epoch": 0.16164715713336575,
      "grad_norm": 0.041547488421201706,
      "learning_rate": 1.3889182081860962e-07,
      "loss": 10.2681,
      "step": 1475
    },
    {
      "epoch": 0.16175674842633753,
      "grad_norm": 0.045067913830280304,
      "learning_rate": 1.2800502578991235e-07,
      "loss": 10.2684,
      "step": 1476
    },
    {
      "epoch": 0.1618663397193093,
      "grad_norm": 0.05577077716588974,
      "learning_rate": 1.1756221829148928e-07,
      "loss": 10.2721,
      "step": 1477
    },
    {
      "epoch": 0.16197593101228108,
      "grad_norm": 0.0451313853263855,
      "learning_rate": 1.0756344474753821e-07,
      "loss": 10.2736,
      "step": 1478
    },
    {
      "epoch": 0.16208552230525286,
      "grad_norm": 0.04688798636198044,
      "learning_rate": 9.800874960826933e-08,
      "loss": 10.2703,
      "step": 1479
    },
    {
      "epoch": 0.1621951135982246,
      "grad_norm": 0.0532434917986393,
      "learning_rate": 8.889817534969425e-08,
      "loss": 10.2828,
      "step": 1480
    },
    {
      "epoch": 0.1623047048911964,
      "grad_norm": 0.05161526799201965,
      "learning_rate": 8.023176247348163e-08,
      "loss": 10.2838,
      "step": 1481
    },
    {
      "epoch": 0.16241429618416817,
      "grad_norm": 0.0507974773645401,
      "learning_rate": 7.200954950673522e-08,
      "loss": 10.2743,
      "step": 1482
    },
    {
      "epoch": 0.16252388747713994,
      "grad_norm": 0.05589978024363518,
      "learning_rate": 6.423157300184946e-08,
      "loss": 10.2765,
      "step": 1483
    },
    {
      "epoch": 0.16263347877011172,
      "grad_norm": 0.05318177118897438,
      "learning_rate": 5.6897867536331864e-08,
      "loss": 10.2745,
      "step": 1484
    },
    {
      "epoch": 0.1627430700630835,
      "grad_norm": 0.057751212269067764,
      "learning_rate": 5.000846571264761e-08,
      "loss": 10.265,
      "step": 1485
    },
    {
      "epoch": 0.16285266135605525,
      "grad_norm": 0.046013571321964264,
      "learning_rate": 4.35633981580974e-08,
      "loss": 10.2773,
      "step": 1486
    },
    {
      "epoch": 0.16296225264902703,
      "grad_norm": 0.05695558339357376,
      "learning_rate": 3.756269352462871e-08,
      "loss": 10.2757,
      "step": 1487
    },
    {
      "epoch": 0.1630718439419988,
      "grad_norm": 0.050595615059137344,
      "learning_rate": 3.20063784888025e-08,
      "loss": 10.2694,
      "step": 1488
    },
    {
      "epoch": 0.16318143523497058,
      "grad_norm": 0.04946747049689293,
      "learning_rate": 2.6894477751548964e-08,
      "loss": 10.267,
      "step": 1489
    },
    {
      "epoch": 0.16329102652794236,
      "grad_norm": 0.055464666336774826,
      "learning_rate": 2.222701403818972e-08,
      "loss": 10.2575,
      "step": 1490
    },
    {
      "epoch": 0.16340061782091414,
      "grad_norm": 0.048841964453458786,
      "learning_rate": 1.8004008098226887e-08,
      "loss": 10.2693,
      "step": 1491
    },
    {
      "epoch": 0.1635102091138859,
      "grad_norm": 0.06157677620649338,
      "learning_rate": 1.4225478705309769e-08,
      "loss": 10.2709,
      "step": 1492
    },
    {
      "epoch": 0.16361980040685767,
      "grad_norm": 0.04942382499575615,
      "learning_rate": 1.0891442657134932e-08,
      "loss": 10.2701,
      "step": 1493
    },
    {
      "epoch": 0.16372939169982945,
      "grad_norm": 0.05516134575009346,
      "learning_rate": 8.001914775401798e-09,
      "loss": 10.2695,
      "step": 1494
    },
    {
      "epoch": 0.16383898299280122,
      "grad_norm": 0.05717114731669426,
      "learning_rate": 5.5569079056794206e-09,
      "loss": 10.2779,
      "step": 1495
    },
    {
      "epoch": 0.163948574285773,
      "grad_norm": 0.05512282997369766,
      "learning_rate": 3.5564329174064824e-09,
      "loss": 10.2699,
      "step": 1496
    },
    {
      "epoch": 0.16405816557874478,
      "grad_norm": 0.051132407039403915,
      "learning_rate": 2.0004987038246824e-09,
      "loss": 10.2653,
      "step": 1497
    },
    {
      "epoch": 0.16416775687171653,
      "grad_norm": 0.06292887032032013,
      "learning_rate": 8.891121819565306e-10,
      "loss": 10.2781,
      "step": 1498
    },
    {
      "epoch": 0.1642773481646883,
      "grad_norm": 0.054577313363552094,
      "learning_rate": 2.2227829252763344e-10,
      "loss": 10.2559,
      "step": 1499
    },
    {
      "epoch": 0.1643869394576601,
      "grad_norm": 0.06058506295084953,
      "learning_rate": 0.0,
      "loss": 10.2746,
      "step": 1500
    },
    {
      "epoch": 0.1643869394576601,
      "eval_loss": 10.272268295288086,
      "eval_runtime": 50.9509,
      "eval_samples_per_second": 301.643,
      "eval_steps_per_second": 150.831,
      "step": 1500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 409070680276992.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}