{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0002813291238162,
  "eval_steps": 500,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001000281329123816,
      "grad_norm": 6.540109634399414,
      "learning_rate": 0.0001999999450590425,
      "loss": 2.6368,
      "step": 1
    },
    {
      "epoch": 0.002000562658247632,
      "grad_norm": 3.004662036895752,
      "learning_rate": 0.00019999978023623033,
      "loss": 2.0338,
      "step": 2
    },
    {
      "epoch": 0.003000843987371448,
      "grad_norm": 2.942652463912964,
      "learning_rate": 0.0001999995055317446,
      "loss": 1.6784,
      "step": 3
    },
    {
      "epoch": 0.004001125316495264,
      "grad_norm": 1.9385607242584229,
      "learning_rate": 0.00019999912094588717,
      "loss": 1.5838,
      "step": 4
    },
    {
      "epoch": 0.005001406645619081,
      "grad_norm": 1.7167695760726929,
      "learning_rate": 0.00019999862647908064,
      "loss": 1.4409,
      "step": 5
    },
    {
      "epoch": 0.006001687974742896,
      "grad_norm": 1.8187777996063232,
      "learning_rate": 0.00019999802213186834,
      "loss": 1.4089,
      "step": 6
    },
    {
      "epoch": 0.007001969303866712,
      "grad_norm": 1.9561201333999634,
      "learning_rate": 0.0001999973079049143,
      "loss": 1.4378,
      "step": 7
    },
    {
      "epoch": 0.008002250632990529,
      "grad_norm": 1.5050302743911743,
      "learning_rate": 0.00019999648379900338,
      "loss": 1.1313,
      "step": 8
    },
    {
      "epoch": 0.009002531962114344,
      "grad_norm": 1.2175908088684082,
      "learning_rate": 0.0001999955498150411,
      "loss": 1.161,
      "step": 9
    },
    {
      "epoch": 0.010002813291238161,
      "grad_norm": 1.0915567874908447,
      "learning_rate": 0.00019999450595405374,
      "loss": 1.1367,
      "step": 10
    },
    {
      "epoch": 0.011003094620361977,
      "grad_norm": 1.0935776233673096,
      "learning_rate": 0.0001999933522171883,
      "loss": 0.9341,
      "step": 11
    },
    {
      "epoch": 0.012003375949485792,
      "grad_norm": 1.1541457176208496,
      "learning_rate": 0.00019999208860571255,
      "loss": 1.0971,
      "step": 12
    },
    {
      "epoch": 0.01300365727860961,
      "grad_norm": 1.0622767210006714,
      "learning_rate": 0.00019999071512101496,
      "loss": 1.0928,
      "step": 13
    },
    {
      "epoch": 0.014003938607733425,
      "grad_norm": 1.0497690439224243,
      "learning_rate": 0.00019998923176460474,
      "loss": 1.0885,
      "step": 14
    },
    {
      "epoch": 0.01500421993685724,
      "grad_norm": 0.9466670155525208,
      "learning_rate": 0.00019998763853811184,
      "loss": 0.9602,
      "step": 15
    },
    {
      "epoch": 0.016004501265981057,
      "grad_norm": 1.1238515377044678,
      "learning_rate": 0.00019998593544328692,
      "loss": 0.9708,
      "step": 16
    },
    {
      "epoch": 0.017004782595104875,
      "grad_norm": 1.2019788026809692,
      "learning_rate": 0.00019998412248200138,
      "loss": 0.9854,
      "step": 17
    },
    {
      "epoch": 0.01800506392422869,
      "grad_norm": 0.8101840019226074,
      "learning_rate": 0.00019998219965624734,
      "loss": 0.9747,
      "step": 18
    },
    {
      "epoch": 0.019005345253352506,
      "grad_norm": 0.7606379985809326,
      "learning_rate": 0.0001999801669681376,
      "loss": 0.9558,
      "step": 19
    },
    {
      "epoch": 0.020005626582476323,
      "grad_norm": 0.7452303767204285,
      "learning_rate": 0.00019997802441990573,
      "loss": 0.8407,
      "step": 20
    },
    {
      "epoch": 0.021005907911600136,
      "grad_norm": 0.810576856136322,
      "learning_rate": 0.00019997577201390606,
      "loss": 0.8321,
      "step": 21
    },
    {
      "epoch": 0.022006189240723954,
      "grad_norm": 0.8486515879631042,
      "learning_rate": 0.00019997340975261353,
      "loss": 0.8239,
      "step": 22
    },
    {
      "epoch": 0.02300647056984777,
      "grad_norm": 1.194822907447815,
      "learning_rate": 0.00019997093763862383,
      "loss": 0.85,
      "step": 23
    },
    {
      "epoch": 0.024006751898971584,
      "grad_norm": 0.8571400046348572,
      "learning_rate": 0.0001999683556746534,
      "loss": 0.8917,
      "step": 24
    },
    {
      "epoch": 0.0250070332280954,
      "grad_norm": 0.7690589427947998,
      "learning_rate": 0.0001999656638635393,
      "loss": 0.9666,
      "step": 25
    },
    {
      "epoch": 0.02600731455721922,
      "grad_norm": 0.9334922432899475,
      "learning_rate": 0.0001999628622082394,
      "loss": 0.8808,
      "step": 26
    },
    {
      "epoch": 0.027007595886343033,
      "grad_norm": 0.9193708896636963,
      "learning_rate": 0.0001999599507118322,
      "loss": 0.9316,
      "step": 27
    },
    {
      "epoch": 0.02800787721546685,
      "grad_norm": 0.9622653126716614,
      "learning_rate": 0.00019995692937751683,
      "loss": 0.9242,
      "step": 28
    },
    {
      "epoch": 0.029008158544590667,
      "grad_norm": 0.76296067237854,
      "learning_rate": 0.0001999537982086133,
      "loss": 0.7989,
      "step": 29
    },
    {
      "epoch": 0.03000843987371448,
      "grad_norm": 0.84053635597229,
      "learning_rate": 0.00019995055720856218,
      "loss": 0.8608,
      "step": 30
    },
    {
      "epoch": 0.031008721202838298,
      "grad_norm": 0.7417949438095093,
      "learning_rate": 0.00019994720638092468,
      "loss": 0.822,
      "step": 31
    },
    {
      "epoch": 0.032009002531962115,
      "grad_norm": 0.8762928247451782,
      "learning_rate": 0.00019994374572938277,
      "loss": 0.9001,
      "step": 32
    },
    {
      "epoch": 0.03300928386108593,
      "grad_norm": 0.6854381561279297,
      "learning_rate": 0.00019994017525773913,
      "loss": 0.7946,
      "step": 33
    },
    {
      "epoch": 0.03400956519020975,
      "grad_norm": 0.6967355012893677,
      "learning_rate": 0.00019993649496991705,
      "loss": 0.8277,
      "step": 34
    },
    {
      "epoch": 0.03500984651933356,
      "grad_norm": 1.393047571182251,
      "learning_rate": 0.00019993270486996046,
      "loss": 0.7741,
      "step": 35
    },
    {
      "epoch": 0.03601012784845738,
      "grad_norm": 0.676814615726471,
      "learning_rate": 0.000199928804962034,
      "loss": 0.7317,
      "step": 36
    },
    {
      "epoch": 0.037010409177581194,
      "grad_norm": 1.0221902132034302,
      "learning_rate": 0.00019992479525042303,
      "loss": 0.8664,
      "step": 37
    },
    {
      "epoch": 0.03801069050670501,
      "grad_norm": 0.5960223078727722,
      "learning_rate": 0.00019992067573953342,
      "loss": 0.8324,
      "step": 38
    },
    {
      "epoch": 0.03901097183582883,
      "grad_norm": 0.6729836463928223,
      "learning_rate": 0.0001999164464338918,
      "loss": 0.7634,
      "step": 39
    },
    {
      "epoch": 0.040011253164952645,
      "grad_norm": 0.802059531211853,
      "learning_rate": 0.0001999121073381454,
      "loss": 0.8515,
      "step": 40
    },
    {
      "epoch": 0.041011534494076456,
      "grad_norm": 1.074352741241455,
      "learning_rate": 0.0001999076584570621,
      "loss": 0.811,
      "step": 41
    },
    {
      "epoch": 0.04201181582320027,
      "grad_norm": 1.1671184301376343,
      "learning_rate": 0.00019990309979553045,
      "loss": 0.8336,
      "step": 42
    },
    {
      "epoch": 0.04301209715232409,
      "grad_norm": 1.1403908729553223,
      "learning_rate": 0.00019989843135855958,
      "loss": 0.8674,
      "step": 43
    },
    {
      "epoch": 0.04401237848144791,
      "grad_norm": 0.9987945556640625,
      "learning_rate": 0.00019989365315127922,
      "loss": 0.8558,
      "step": 44
    },
    {
      "epoch": 0.045012659810571724,
      "grad_norm": 0.827928900718689,
      "learning_rate": 0.0001998887651789398,
      "loss": 0.7749,
      "step": 45
    },
    {
      "epoch": 0.04601294113969554,
      "grad_norm": 0.694059431552887,
      "learning_rate": 0.0001998837674469123,
      "loss": 0.8992,
      "step": 46
    },
    {
      "epoch": 0.04701322246881936,
      "grad_norm": 0.5862646102905273,
      "learning_rate": 0.00019987865996068833,
      "loss": 0.7361,
      "step": 47
    },
    {
      "epoch": 0.04801350379794317,
      "grad_norm": 0.5515244007110596,
      "learning_rate": 0.00019987344272588006,
      "loss": 0.7065,
      "step": 48
    },
    {
      "epoch": 0.049013785127066986,
      "grad_norm": 0.5773719549179077,
      "learning_rate": 0.00019986811574822033,
      "loss": 0.7041,
      "step": 49
    },
    {
      "epoch": 0.0500140664561908,
      "grad_norm": 0.4614185094833374,
      "learning_rate": 0.00019986267903356254,
      "loss": 0.6977,
      "step": 50
    },
    {
      "epoch": 0.05101434778531462,
      "grad_norm": 0.5293365716934204,
      "learning_rate": 0.0001998571325878806,
      "loss": 0.7435,
      "step": 51
    },
    {
      "epoch": 0.05201462911443844,
      "grad_norm": 0.4691166281700134,
      "learning_rate": 0.0001998514764172691,
      "loss": 0.684,
      "step": 52
    },
    {
      "epoch": 0.053014910443562255,
      "grad_norm": 0.6643468141555786,
      "learning_rate": 0.00019984571052794313,
      "loss": 0.7708,
      "step": 53
    },
    {
      "epoch": 0.054015191772686065,
      "grad_norm": 0.6328372955322266,
      "learning_rate": 0.00019983983492623833,
      "loss": 0.8069,
      "step": 54
    },
    {
      "epoch": 0.05501547310180988,
      "grad_norm": 0.7191805243492126,
      "learning_rate": 0.00019983384961861096,
      "loss": 0.757,
      "step": 55
    },
    {
      "epoch": 0.0560157544309337,
      "grad_norm": 0.6610512137413025,
      "learning_rate": 0.0001998277546116378,
      "loss": 0.6549,
      "step": 56
    },
    {
      "epoch": 0.05701603576005752,
      "grad_norm": 0.6019318699836731,
      "learning_rate": 0.00019982154991201608,
      "loss": 0.7135,
      "step": 57
    },
    {
      "epoch": 0.058016317089181334,
      "grad_norm": 0.6172312498092651,
      "learning_rate": 0.00019981523552656377,
      "loss": 0.7454,
      "step": 58
    },
    {
      "epoch": 0.05901659841830515,
      "grad_norm": 0.6532041430473328,
      "learning_rate": 0.00019980881146221914,
      "loss": 0.6976,
      "step": 59
    },
    {
      "epoch": 0.06001687974742896,
      "grad_norm": 0.7116113305091858,
      "learning_rate": 0.00019980227772604112,
      "loss": 0.7581,
      "step": 60
    },
    {
      "epoch": 0.06101716107655278,
      "grad_norm": 0.6532421708106995,
      "learning_rate": 0.0001997956343252091,
      "loss": 0.7042,
      "step": 61
    },
    {
      "epoch": 0.062017442405676595,
      "grad_norm": 0.6296349167823792,
      "learning_rate": 0.00019978888126702296,
      "loss": 0.6981,
      "step": 62
    },
    {
      "epoch": 0.06301772373480041,
      "grad_norm": 0.6616079211235046,
      "learning_rate": 0.00019978201855890308,
      "loss": 0.6885,
      "step": 63
    },
    {
      "epoch": 0.06401800506392423,
      "grad_norm": 0.6802906394004822,
      "learning_rate": 0.00019977504620839035,
      "loss": 0.6685,
      "step": 64
    },
    {
      "epoch": 0.06501828639304805,
      "grad_norm": 0.5653461217880249,
      "learning_rate": 0.00019976796422314615,
      "loss": 0.7135,
      "step": 65
    },
    {
      "epoch": 0.06601856772217186,
      "grad_norm": 0.5990989804267883,
      "learning_rate": 0.00019976077261095226,
      "loss": 0.8312,
      "step": 66
    },
    {
      "epoch": 0.06701884905129568,
      "grad_norm": 0.6948749423027039,
      "learning_rate": 0.00019975347137971098,
      "loss": 0.742,
      "step": 67
    },
    {
      "epoch": 0.0680191303804195,
      "grad_norm": 0.6895416378974915,
      "learning_rate": 0.00019974606053744503,
      "loss": 0.6902,
      "step": 68
    },
    {
      "epoch": 0.06901941170954332,
      "grad_norm": 0.6278745532035828,
      "learning_rate": 0.00019973854009229763,
      "loss": 0.65,
      "step": 69
    },
    {
      "epoch": 0.07001969303866712,
      "grad_norm": 0.6024957299232483,
      "learning_rate": 0.00019973091005253232,
      "loss": 0.6611,
      "step": 70
    },
    {
      "epoch": 0.07101997436779094,
      "grad_norm": 1.4569993019104004,
      "learning_rate": 0.0001997231704265332,
      "loss": 0.6809,
      "step": 71
    },
    {
      "epoch": 0.07202025569691475,
      "grad_norm": 0.42510971426963806,
      "learning_rate": 0.00019971532122280464,
      "loss": 0.6415,
      "step": 72
    },
    {
      "epoch": 0.07302053702603857,
      "grad_norm": 0.5024048089981079,
      "learning_rate": 0.0001997073624499716,
      "loss": 0.6515,
      "step": 73
    },
    {
      "epoch": 0.07402081835516239,
      "grad_norm": 0.4638510048389435,
      "learning_rate": 0.0001996992941167792,
      "loss": 0.726,
      "step": 74
    },
    {
      "epoch": 0.0750210996842862,
      "grad_norm": 0.5222762823104858,
      "learning_rate": 0.00019969111623209323,
      "loss": 0.5953,
      "step": 75
    },
    {
      "epoch": 0.07602138101341002,
      "grad_norm": 0.5868512392044067,
      "learning_rate": 0.00019968282880489957,
      "loss": 0.7317,
      "step": 76
    },
    {
      "epoch": 0.07702166234253384,
      "grad_norm": 0.6825968027114868,
      "learning_rate": 0.00019967443184430467,
      "loss": 0.6028,
      "step": 77
    },
    {
      "epoch": 0.07802194367165766,
      "grad_norm": 0.7539929747581482,
      "learning_rate": 0.0001996659253595353,
      "loss": 0.747,
      "step": 78
    },
    {
      "epoch": 0.07902222500078147,
      "grad_norm": 0.7608583569526672,
      "learning_rate": 0.0001996573093599385,
      "loss": 0.6783,
      "step": 79
    },
    {
      "epoch": 0.08002250632990529,
      "grad_norm": 0.7046730518341064,
      "learning_rate": 0.00019964858385498172,
      "loss": 0.6443,
      "step": 80
    },
    {
      "epoch": 0.08102278765902911,
      "grad_norm": 0.7782636284828186,
      "learning_rate": 0.00019963974885425266,
      "loss": 0.6666,
      "step": 81
    },
    {
      "epoch": 0.08202306898815291,
      "grad_norm": 0.7668784856796265,
      "learning_rate": 0.00019963080436745945,
      "loss": 0.6789,
      "step": 82
    },
    {
      "epoch": 0.08302335031727673,
      "grad_norm": 0.759543240070343,
      "learning_rate": 0.00019962175040443044,
      "loss": 0.7122,
      "step": 83
    },
    {
      "epoch": 0.08402363164640055,
      "grad_norm": 0.4773256778717041,
      "learning_rate": 0.0001996125869751143,
      "loss": 0.61,
      "step": 84
    },
    {
      "epoch": 0.08502391297552436,
      "grad_norm": 0.4939149022102356,
      "learning_rate": 0.00019960331408957997,
      "loss": 0.6302,
      "step": 85
    },
    {
      "epoch": 0.08602419430464818,
      "grad_norm": 0.5174906849861145,
      "learning_rate": 0.00019959393175801671,
      "loss": 0.682,
      "step": 86
    },
    {
      "epoch": 0.087024475633772,
      "grad_norm": 0.4596863090991974,
      "learning_rate": 0.00019958443999073397,
      "loss": 0.6479,
      "step": 87
    },
    {
      "epoch": 0.08802475696289581,
      "grad_norm": 0.5851575136184692,
      "learning_rate": 0.00019957483879816151,
      "loss": 0.6814,
      "step": 88
    },
    {
      "epoch": 0.08902503829201963,
      "grad_norm": 0.5562273859977722,
      "learning_rate": 0.00019956512819084928,
      "loss": 0.6806,
      "step": 89
    },
    {
      "epoch": 0.09002531962114345,
      "grad_norm": 0.5329018235206604,
      "learning_rate": 0.00019955530817946748,
      "loss": 0.6147,
      "step": 90
    },
    {
      "epoch": 0.09102560095026727,
      "grad_norm": 0.5893882513046265,
      "learning_rate": 0.00019954537877480655,
      "loss": 0.5858,
      "step": 91
    },
    {
      "epoch": 0.09202588227939108,
      "grad_norm": 0.5248862504959106,
      "learning_rate": 0.00019953533998777706,
      "loss": 0.602,
      "step": 92
    },
    {
      "epoch": 0.0930261636085149,
      "grad_norm": 0.5627758502960205,
      "learning_rate": 0.00019952519182940993,
      "loss": 0.6584,
      "step": 93
    },
    {
      "epoch": 0.09402644493763872,
      "grad_norm": 0.6081700325012207,
      "learning_rate": 0.00019951493431085603,
      "loss": 0.6227,
      "step": 94
    },
    {
      "epoch": 0.09502672626676252,
      "grad_norm": 0.6908998489379883,
      "learning_rate": 0.00019950456744338658,
      "loss": 0.6615,
      "step": 95
    },
    {
      "epoch": 0.09602700759588634,
      "grad_norm": 0.49264904856681824,
      "learning_rate": 0.00019949409123839288,
      "loss": 0.6784,
      "step": 96
    },
    {
      "epoch": 0.09702728892501016,
      "grad_norm": 0.6333140730857849,
      "learning_rate": 0.00019948350570738642,
      "loss": 0.661,
      "step": 97
    },
    {
      "epoch": 0.09802757025413397,
      "grad_norm": 0.6504009962081909,
      "learning_rate": 0.0001994728108619987,
      "loss": 0.649,
      "step": 98
    },
    {
      "epoch": 0.09902785158325779,
      "grad_norm": 0.6558539867401123,
      "learning_rate": 0.0001994620067139815,
      "loss": 0.5927,
      "step": 99
    },
    {
      "epoch": 0.1000281329123816,
      "grad_norm": 0.750813901424408,
      "learning_rate": 0.00019945109327520658,
      "loss": 0.6479,
      "step": 100
    },
    {
      "epoch": 0.10102841424150542,
      "grad_norm": 0.6053513884544373,
      "learning_rate": 0.00019944007055766586,
      "loss": 0.6851,
      "step": 101
    },
    {
      "epoch": 0.10202869557062924,
      "grad_norm": 0.5293607711791992,
      "learning_rate": 0.00019942893857347128,
      "loss": 0.6544,
      "step": 102
    },
    {
      "epoch": 0.10302897689975306,
      "grad_norm": 0.5119344592094421,
      "learning_rate": 0.00019941769733485494,
      "loss": 0.6704,
      "step": 103
    },
    {
      "epoch": 0.10402925822887688,
      "grad_norm": 0.49486231803894043,
      "learning_rate": 0.00019940634685416888,
      "loss": 0.6309,
      "step": 104
    },
    {
      "epoch": 0.10502953955800069,
      "grad_norm": 0.408372700214386,
      "learning_rate": 0.00019939488714388524,
      "loss": 0.5601,
      "step": 105
    },
    {
      "epoch": 0.10602982088712451,
      "grad_norm": 0.45692771673202515,
      "learning_rate": 0.00019938331821659614,
      "loss": 0.583,
      "step": 106
    },
    {
      "epoch": 0.10703010221624833,
      "grad_norm": 0.5024964809417725,
      "learning_rate": 0.0001993716400850138,
      "loss": 0.5944,
      "step": 107
    },
    {
      "epoch": 0.10803038354537213,
      "grad_norm": 0.621824324131012,
      "learning_rate": 0.0001993598527619703,
      "loss": 0.6284,
      "step": 108
    },
    {
      "epoch": 0.10903066487449595,
      "grad_norm": 0.5267654657363892,
      "learning_rate": 0.00019934795626041783,
      "loss": 0.6178,
      "step": 109
    },
    {
      "epoch": 0.11003094620361976,
      "grad_norm": 0.497410386800766,
      "learning_rate": 0.0001993359505934285,
      "loss": 0.557,
      "step": 110
    },
    {
      "epoch": 0.11103122753274358,
      "grad_norm": 0.5206043720245361,
      "learning_rate": 0.00019932383577419432,
      "loss": 0.577,
      "step": 111
    },
    {
      "epoch": 0.1120315088618674,
      "grad_norm": 1.2140614986419678,
      "learning_rate": 0.0001993116118160273,
      "loss": 0.6339,
      "step": 112
    },
    {
      "epoch": 0.11303179019099122,
      "grad_norm": 0.7585737109184265,
      "learning_rate": 0.00019929927873235938,
      "loss": 0.6587,
      "step": 113
    },
    {
      "epoch": 0.11403207152011503,
      "grad_norm": 0.5759165287017822,
      "learning_rate": 0.00019928683653674237,
      "loss": 0.6275,
      "step": 114
    },
    {
      "epoch": 0.11503235284923885,
      "grad_norm": 0.5694018006324768,
      "learning_rate": 0.00019927428524284805,
      "loss": 0.6535,
      "step": 115
    },
    {
      "epoch": 0.11603263417836267,
      "grad_norm": 0.4953472912311554,
      "learning_rate": 0.00019926162486446792,
      "loss": 0.6077,
      "step": 116
    },
    {
      "epoch": 0.11703291550748648,
      "grad_norm": 0.6509867310523987,
      "learning_rate": 0.0001992488554155135,
      "loss": 0.6451,
      "step": 117
    },
    {
      "epoch": 0.1180331968366103,
      "grad_norm": 0.6137562990188599,
      "learning_rate": 0.00019923597691001615,
      "loss": 0.6224,
      "step": 118
    },
    {
      "epoch": 0.11903347816573412,
      "grad_norm": 0.595355212688446,
      "learning_rate": 0.0001992229893621269,
      "loss": 0.6059,
      "step": 119
    },
    {
      "epoch": 0.12003375949485792,
      "grad_norm": 0.5127987265586853,
      "learning_rate": 0.00019920989278611687,
      "loss": 0.6359,
      "step": 120
    },
    {
      "epoch": 0.12103404082398174,
      "grad_norm": 0.5433658361434937,
      "learning_rate": 0.0001991966871963767,
      "loss": 0.6817,
      "step": 121
    },
    {
      "epoch": 0.12203432215310556,
      "grad_norm": 0.634130597114563,
      "learning_rate": 0.000199183372607417,
      "loss": 0.6174,
      "step": 122
    },
    {
      "epoch": 0.12303460348222937,
      "grad_norm": 0.5337157845497131,
      "learning_rate": 0.0001991699490338681,
      "loss": 0.671,
      "step": 123
    },
    {
      "epoch": 0.12403488481135319,
      "grad_norm": 0.5023785829544067,
      "learning_rate": 0.00019915641649048005,
      "loss": 0.5828,
      "step": 124
    },
    {
      "epoch": 0.12503516614047702,
      "grad_norm": 0.711187481880188,
      "learning_rate": 0.0001991427749921227,
      "loss": 0.5384,
      "step": 125
    },
    {
      "epoch": 0.12603544746960083,
      "grad_norm": 0.7347482442855835,
      "learning_rate": 0.00019912902455378556,
      "loss": 0.6257,
      "step": 126
    },
    {
      "epoch": 0.12703572879872463,
      "grad_norm": 0.583920955657959,
      "learning_rate": 0.00019911516519057788,
      "loss": 0.5547,
      "step": 127
    },
    {
      "epoch": 0.12803601012784846,
      "grad_norm": 1.15117609500885,
      "learning_rate": 0.00019910119691772863,
      "loss": 0.6215,
      "step": 128
    },
    {
      "epoch": 0.12903629145697226,
      "grad_norm": 0.5940455794334412,
      "learning_rate": 0.00019908711975058637,
      "loss": 0.6136,
      "step": 129
    },
    {
      "epoch": 0.1300365727860961,
      "grad_norm": 0.5960786938667297,
      "learning_rate": 0.0001990729337046194,
      "loss": 0.5271,
      "step": 130
    },
    {
      "epoch": 0.1310368541152199,
      "grad_norm": 0.6204031705856323,
      "learning_rate": 0.0001990586387954156,
      "loss": 0.5584,
      "step": 131
    },
    {
      "epoch": 0.13203713544434373,
      "grad_norm": 0.6092723608016968,
      "learning_rate": 0.00019904423503868247,
      "loss": 0.6402,
      "step": 132
    },
    {
      "epoch": 0.13303741677346753,
      "grad_norm": 0.6005457639694214,
      "learning_rate": 0.00019902972245024715,
      "loss": 0.6422,
      "step": 133
    },
    {
      "epoch": 0.13403769810259136,
      "grad_norm": 0.6619305610656738,
      "learning_rate": 0.00019901510104605637,
      "loss": 0.5395,
      "step": 134
    },
    {
      "epoch": 0.13503797943171517,
      "grad_norm": 0.7267995476722717,
      "learning_rate": 0.00019900037084217637,
      "loss": 0.6358,
      "step": 135
    },
    {
      "epoch": 0.136038260760839,
      "grad_norm": 0.3749965727329254,
      "learning_rate": 0.00019898553185479303,
      "loss": 0.5185,
      "step": 136
    },
    {
      "epoch": 0.1370385420899628,
      "grad_norm": 0.6056541204452515,
      "learning_rate": 0.00019897058410021167,
      "loss": 0.5466,
      "step": 137
    },
    {
      "epoch": 0.13803882341908663,
      "grad_norm": 0.4612765908241272,
      "learning_rate": 0.00019895552759485722,
      "loss": 0.6349,
      "step": 138
    },
    {
      "epoch": 0.13903910474821043,
      "grad_norm": 0.47530749440193176,
      "learning_rate": 0.00019894036235527395,
      "loss": 0.5828,
      "step": 139
    },
    {
      "epoch": 0.14003938607733424,
      "grad_norm": 0.5521253347396851,
      "learning_rate": 0.00019892508839812584,
      "loss": 0.6172,
      "step": 140
    },
    {
      "epoch": 0.14103966740645807,
      "grad_norm": 0.4842628538608551,
      "learning_rate": 0.00019890970574019617,
      "loss": 0.5327,
      "step": 141
    },
    {
      "epoch": 0.14203994873558187,
      "grad_norm": 0.5947982668876648,
      "learning_rate": 0.00019889421439838763,
      "loss": 0.5432,
      "step": 142
    },
    {
      "epoch": 0.1430402300647057,
      "grad_norm": 0.6485494375228882,
      "learning_rate": 0.00019887861438972246,
      "loss": 0.5293,
      "step": 143
    },
    {
      "epoch": 0.1440405113938295,
      "grad_norm": 0.4645004868507385,
      "learning_rate": 0.00019886290573134228,
      "loss": 0.6078,
      "step": 144
    },
    {
      "epoch": 0.14504079272295334,
      "grad_norm": 0.49198296666145325,
      "learning_rate": 0.000198847088440508,
      "loss": 0.5283,
      "step": 145
    },
    {
      "epoch": 0.14604107405207714,
      "grad_norm": 0.6396425366401672,
      "learning_rate": 0.0001988311625346,
      "loss": 0.6178,
      "step": 146
    },
    {
      "epoch": 0.14704135538120097,
      "grad_norm": 0.49430540204048157,
      "learning_rate": 0.00019881512803111796,
      "loss": 0.5914,
      "step": 147
    },
    {
      "epoch": 0.14804163671032478,
      "grad_norm": 0.5589768886566162,
      "learning_rate": 0.00019879898494768093,
      "loss": 0.633,
      "step": 148
    },
    {
      "epoch": 0.1490419180394486,
      "grad_norm": 0.629703938961029,
      "learning_rate": 0.00019878273330202717,
      "loss": 0.5706,
      "step": 149
    },
    {
      "epoch": 0.1500421993685724,
      "grad_norm": 0.4935409724712372,
      "learning_rate": 0.00019876637311201433,
      "loss": 0.5004,
      "step": 150
    },
    {
      "epoch": 0.15104248069769624,
      "grad_norm": 0.44915804266929626,
      "learning_rate": 0.00019874990439561934,
      "loss": 0.5326,
      "step": 151
    },
    {
      "epoch": 0.15204276202682004,
      "grad_norm": 0.5451924204826355,
      "learning_rate": 0.0001987333271709383,
      "loss": 0.6474,
      "step": 152
    },
    {
      "epoch": 0.15304304335594385,
      "grad_norm": 0.641677737236023,
      "learning_rate": 0.00019871664145618657,
      "loss": 0.5785,
      "step": 153
    },
    {
      "epoch": 0.15404332468506768,
      "grad_norm": 0.40758514404296875,
      "learning_rate": 0.00019869984726969878,
      "loss": 0.5416,
      "step": 154
    },
    {
      "epoch": 0.15504360601419148,
      "grad_norm": 0.8472537994384766,
      "learning_rate": 0.00019868294462992866,
      "loss": 0.5764,
      "step": 155
    },
    {
      "epoch": 0.1560438873433153,
      "grad_norm": 0.49797093868255615,
      "learning_rate": 0.00019866593355544922,
      "loss": 0.5597,
      "step": 156
    },
    {
      "epoch": 0.15704416867243912,
      "grad_norm": 0.5156570076942444,
      "learning_rate": 0.00019864881406495246,
      "loss": 0.6066,
      "step": 157
    },
    {
      "epoch": 0.15804445000156295,
      "grad_norm": 0.5762325525283813,
      "learning_rate": 0.00019863158617724967,
      "loss": 0.5678,
      "step": 158
    },
    {
      "epoch": 0.15904473133068675,
      "grad_norm": 0.4978669583797455,
      "learning_rate": 0.00019861424991127115,
      "loss": 0.5523,
      "step": 159
    },
    {
      "epoch": 0.16004501265981058,
      "grad_norm": 1.5707205533981323,
      "learning_rate": 0.00019859680528606637,
      "loss": 0.6809,
      "step": 160
    },
    {
      "epoch": 0.16104529398893438,
      "grad_norm": 0.42276397347450256,
      "learning_rate": 0.00019857925232080373,
      "loss": 0.5389,
      "step": 161
    },
    {
      "epoch": 0.16204557531805822,
      "grad_norm": 0.5103838443756104,
      "learning_rate": 0.00019856159103477086,
      "loss": 0.5572,
      "step": 162
    },
    {
      "epoch": 0.16304585664718202,
      "grad_norm": 0.49496352672576904,
      "learning_rate": 0.00019854382144737426,
      "loss": 0.5608,
      "step": 163
    },
    {
      "epoch": 0.16404613797630582,
      "grad_norm": 0.48654571175575256,
      "learning_rate": 0.00019852594357813952,
      "loss": 0.5678,
      "step": 164
    },
    {
      "epoch": 0.16504641930542965,
      "grad_norm": 0.6116751432418823,
      "learning_rate": 0.00019850795744671116,
      "loss": 0.6197,
      "step": 165
    },
    {
      "epoch": 0.16604670063455346,
      "grad_norm": 0.4954712688922882,
      "learning_rate": 0.0001984898630728527,
      "loss": 0.5498,
      "step": 166
    },
    {
      "epoch": 0.1670469819636773,
      "grad_norm": 0.5408039689064026,
      "learning_rate": 0.0001984716604764466,
      "loss": 0.5202,
      "step": 167
    },
    {
      "epoch": 0.1680472632928011,
      "grad_norm": 0.6083703637123108,
      "learning_rate": 0.0001984533496774942,
      "loss": 0.5584,
      "step": 168
    },
    {
      "epoch": 0.16904754462192492,
      "grad_norm": 0.47014451026916504,
      "learning_rate": 0.0001984349306961158,
      "loss": 0.5508,
      "step": 169
    },
    {
      "epoch": 0.17004782595104873,
      "grad_norm": 0.5064229369163513,
      "learning_rate": 0.00019841640355255043,
      "loss": 0.6152,
      "step": 170
    },
    {
      "epoch": 0.17104810728017256,
      "grad_norm": 0.6776212453842163,
      "learning_rate": 0.00019839776826715614,
      "loss": 0.5392,
      "step": 171
    },
    {
      "epoch": 0.17204838860929636,
      "grad_norm": 0.46285948157310486,
      "learning_rate": 0.00019837902486040978,
      "loss": 0.6151,
      "step": 172
    },
    {
      "epoch": 0.1730486699384202,
      "grad_norm": 0.48528385162353516,
      "learning_rate": 0.0001983601733529069,
      "loss": 0.5609,
      "step": 173
    },
    {
      "epoch": 0.174048951267544,
      "grad_norm": 0.5141404867172241,
      "learning_rate": 0.00019834121376536187,
      "loss": 0.6115,
      "step": 174
    },
    {
      "epoch": 0.17504923259666783,
      "grad_norm": 0.42379871010780334,
      "learning_rate": 0.00019832214611860793,
      "loss": 0.6509,
      "step": 175
    },
    {
      "epoch": 0.17604951392579163,
      "grad_norm": 0.4130716323852539,
      "learning_rate": 0.00019830297043359692,
      "loss": 0.5719,
      "step": 176
    },
    {
      "epoch": 0.17704979525491543,
      "grad_norm": 0.41237375140190125,
      "learning_rate": 0.00019828368673139947,
      "loss": 0.5288,
      "step": 177
    },
    {
      "epoch": 0.17805007658403926,
      "grad_norm": 0.5197924971580505,
      "learning_rate": 0.0001982642950332049,
      "loss": 0.5941,
      "step": 178
    },
    {
      "epoch": 0.17905035791316307,
      "grad_norm": 0.4431571066379547,
      "learning_rate": 0.00019824479536032112,
      "loss": 0.5075,
      "step": 179
    },
    {
      "epoch": 0.1800506392422869,
      "grad_norm": 0.5234508514404297,
      "learning_rate": 0.0001982251877341748,
      "loss": 0.5239,
      "step": 180
    },
    {
      "epoch": 0.1810509205714107,
      "grad_norm": 0.5809316635131836,
      "learning_rate": 0.00019820547217631117,
      "loss": 0.5646,
      "step": 181
    },
    {
      "epoch": 0.18205120190053453,
      "grad_norm": 0.5900911092758179,
      "learning_rate": 0.00019818564870839405,
      "loss": 0.5413,
      "step": 182
    },
    {
      "epoch": 0.18305148322965833,
      "grad_norm": 0.6365839242935181,
      "learning_rate": 0.00019816571735220583,
      "loss": 0.5463,
      "step": 183
    },
    {
      "epoch": 0.18405176455878217,
      "grad_norm": 0.7070748805999756,
      "learning_rate": 0.00019814567812964748,
      "loss": 0.5335,
      "step": 184
    },
    {
      "epoch": 0.18505204588790597,
      "grad_norm": 0.5305525064468384,
      "learning_rate": 0.00019812553106273847,
      "loss": 0.529,
      "step": 185
    },
    {
      "epoch": 0.1860523272170298,
      "grad_norm": 0.6620722413063049,
      "learning_rate": 0.00019810527617361681,
      "loss": 0.498,
      "step": 186
    },
    {
      "epoch": 0.1870526085461536,
      "grad_norm": 0.5616530776023865,
      "learning_rate": 0.00019808491348453894,
      "loss": 0.5551,
      "step": 187
    },
    {
      "epoch": 0.18805288987527743,
      "grad_norm": 0.6696614623069763,
      "learning_rate": 0.00019806444301787978,
      "loss": 0.5957,
      "step": 188
    },
    {
      "epoch": 0.18905317120440124,
      "grad_norm": 0.7320078015327454,
      "learning_rate": 0.0001980438647961327,
      "loss": 0.6154,
      "step": 189
    },
    {
      "epoch": 0.19005345253352504,
      "grad_norm": 0.630651593208313,
      "learning_rate": 0.00019802317884190935,
      "loss": 0.5962,
      "step": 190
    },
    {
      "epoch": 0.19105373386264887,
      "grad_norm": 0.6430137753486633,
      "learning_rate": 0.00019800238517793996,
      "loss": 0.5237,
      "step": 191
    },
    {
      "epoch": 0.19205401519177268,
      "grad_norm": 0.5368555188179016,
      "learning_rate": 0.00019798148382707296,
      "loss": 0.5346,
      "step": 192
    },
    {
      "epoch": 0.1930542965208965,
      "grad_norm": 0.45558878779411316,
      "learning_rate": 0.00019796047481227515,
      "loss": 0.567,
      "step": 193
    },
    {
      "epoch": 0.1940545778500203,
      "grad_norm": 0.49816155433654785,
      "learning_rate": 0.00019793935815663163,
      "loss": 0.5741,
      "step": 194
    },
    {
      "epoch": 0.19505485917914414,
      "grad_norm": 0.5015010833740234,
      "learning_rate": 0.00019791813388334581,
      "loss": 0.5198,
      "step": 195
    },
    {
      "epoch": 0.19605514050826794,
      "grad_norm": 0.41044294834136963,
      "learning_rate": 0.00019789680201573933,
      "loss": 0.5111,
      "step": 196
    },
    {
      "epoch": 0.19705542183739178,
      "grad_norm": 0.9961733222007751,
      "learning_rate": 0.00019787536257725202,
      "loss": 0.5427,
      "step": 197
    },
    {
      "epoch": 0.19805570316651558,
      "grad_norm": 0.42929476499557495,
      "learning_rate": 0.00019785381559144196,
      "loss": 0.494,
      "step": 198
    },
    {
      "epoch": 0.1990559844956394,
      "grad_norm": 0.4490405321121216,
      "learning_rate": 0.00019783216108198542,
      "loss": 0.5225,
      "step": 199
    },
    {
      "epoch": 0.2000562658247632,
      "grad_norm": 0.513647198677063,
      "learning_rate": 0.00019781039907267677,
      "loss": 0.557,
      "step": 200
    },
    {
      "epoch": 0.20105654715388704,
      "grad_norm": 0.4340246915817261,
      "learning_rate": 0.00019778852958742853,
      "loss": 0.572,
      "step": 201
    },
    {
      "epoch": 0.20205682848301085,
      "grad_norm": 0.4050307273864746,
      "learning_rate": 0.00019776655265027127,
      "loss": 0.5278,
      "step": 202
    },
    {
      "epoch": 0.20305710981213465,
      "grad_norm": 0.4073551595211029,
      "learning_rate": 0.00019774446828535371,
      "loss": 0.4888,
      "step": 203
    },
    {
      "epoch": 0.20405739114125848,
      "grad_norm": 0.5057980418205261,
      "learning_rate": 0.00019772227651694256,
      "loss": 0.5399,
      "step": 204
    },
    {
      "epoch": 0.20505767247038229,
      "grad_norm": 0.4432837963104248,
      "learning_rate": 0.00019769997736942258,
      "loss": 0.5073,
      "step": 205
    },
    {
      "epoch": 0.20605795379950612,
      "grad_norm": 0.47015416622161865,
      "learning_rate": 0.00019767757086729647,
      "loss": 0.4948,
      "step": 206
    },
    {
      "epoch": 0.20705823512862992,
      "grad_norm": 0.5023201107978821,
      "learning_rate": 0.00019765505703518496,
      "loss": 0.5286,
      "step": 207
    },
    {
      "epoch": 0.20805851645775375,
      "grad_norm": 0.5784673690795898,
      "learning_rate": 0.00019763243589782662,
      "loss": 0.5247,
      "step": 208
    },
    {
      "epoch": 0.20905879778687755,
      "grad_norm": 0.46152037382125854,
      "learning_rate": 0.00019760970748007803,
      "loss": 0.5048,
      "step": 209
    },
    {
      "epoch": 0.21005907911600138,
      "grad_norm": 0.6008761525154114,
      "learning_rate": 0.0001975868718069136,
      "loss": 0.5575,
      "step": 210
    },
    {
      "epoch": 0.2110593604451252,
      "grad_norm": 0.4951910972595215,
      "learning_rate": 0.00019756392890342563,
      "loss": 0.4928,
      "step": 211
    },
    {
      "epoch": 0.21205964177424902,
      "grad_norm": 0.5855419635772705,
      "learning_rate": 0.00019754087879482422,
      "loss": 0.542,
      "step": 212
    },
    {
      "epoch": 0.21305992310337282,
      "grad_norm": 0.5546546578407288,
      "learning_rate": 0.00019751772150643722,
      "loss": 0.5762,
      "step": 213
    },
    {
      "epoch": 0.21406020443249665,
      "grad_norm": 0.5548007488250732,
      "learning_rate": 0.00019749445706371038,
      "loss": 0.522,
      "step": 214
    },
    {
      "epoch": 0.21506048576162046,
      "grad_norm": 0.5254936814308167,
      "learning_rate": 0.00019747108549220702,
      "loss": 0.5864,
      "step": 215
    },
    {
      "epoch": 0.21606076709074426,
      "grad_norm": 0.5107640027999878,
      "learning_rate": 0.00019744760681760832,
      "loss": 0.5451,
      "step": 216
    },
    {
      "epoch": 0.2170610484198681,
      "grad_norm": 0.5446146726608276,
      "learning_rate": 0.00019742402106571314,
      "loss": 0.5352,
      "step": 217
    },
    {
      "epoch": 0.2180613297489919,
      "grad_norm": 0.414689302444458,
      "learning_rate": 0.00019740032826243788,
      "loss": 0.5236,
      "step": 218
    },
    {
      "epoch": 0.21906161107811573,
      "grad_norm": 0.40488916635513306,
      "learning_rate": 0.0001973765284338167,
      "loss": 0.4937,
      "step": 219
    },
    {
      "epoch": 0.22006189240723953,
      "grad_norm": 0.42548874020576477,
      "learning_rate": 0.00019735262160600127,
      "loss": 0.5433,
      "step": 220
    },
    {
      "epoch": 0.22106217373636336,
      "grad_norm": 0.5061275362968445,
      "learning_rate": 0.00019732860780526088,
      "loss": 0.5747,
      "step": 221
    },
    {
      "epoch": 0.22206245506548716,
      "grad_norm": 0.4437318742275238,
      "learning_rate": 0.00019730448705798239,
      "loss": 0.5681,
      "step": 222
    },
    {
      "epoch": 0.223062736394611,
      "grad_norm": 0.40250784158706665,
      "learning_rate": 0.00019728025939067008,
      "loss": 0.4735,
      "step": 223
    },
    {
      "epoch": 0.2240630177237348,
      "grad_norm": 0.43195122480392456,
      "learning_rate": 0.00019725592482994583,
      "loss": 0.5166,
      "step": 224
    },
    {
      "epoch": 0.22506329905285863,
      "grad_norm": 0.3928036689758301,
      "learning_rate": 0.00019723148340254892,
      "loss": 0.4718,
      "step": 225
    },
    {
      "epoch": 0.22606358038198243,
      "grad_norm": 0.3949666917324066,
      "learning_rate": 0.00019720693513533598,
      "loss": 0.4851,
      "step": 226
    },
    {
      "epoch": 0.22706386171110624,
      "grad_norm": 0.4328802227973938,
      "learning_rate": 0.00019718228005528122,
      "loss": 0.4482,
      "step": 227
    },
    {
      "epoch": 0.22806414304023007,
      "grad_norm": 0.5324240326881409,
      "learning_rate": 0.00019715751818947603,
      "loss": 0.5488,
      "step": 228
    },
    {
      "epoch": 0.22906442436935387,
      "grad_norm": 0.446676641702652,
      "learning_rate": 0.0001971326495651293,
      "loss": 0.5178,
      "step": 229
    },
    {
      "epoch": 0.2300647056984777,
      "grad_norm": 0.5981528162956238,
      "learning_rate": 0.00019710767420956705,
      "loss": 0.4943,
      "step": 230
    },
    {
      "epoch": 0.2310649870276015,
      "grad_norm": 0.4626360237598419,
      "learning_rate": 0.0001970825921502328,
      "loss": 0.5424,
      "step": 231
    },
    {
      "epoch": 0.23206526835672533,
      "grad_norm": 0.4125116467475891,
      "learning_rate": 0.0001970574034146871,
      "loss": 0.5126,
      "step": 232
    },
    {
      "epoch": 0.23306554968584914,
      "grad_norm": 0.45425575971603394,
      "learning_rate": 0.00019703210803060782,
      "loss": 0.5083,
      "step": 233
    },
    {
      "epoch": 0.23406583101497297,
      "grad_norm": 0.449907511472702,
      "learning_rate": 0.00019700670602579008,
      "loss": 0.5083,
      "step": 234
    },
    {
      "epoch": 0.23506611234409677,
      "grad_norm": 0.42983028292655945,
      "learning_rate": 0.00019698119742814606,
      "loss": 0.5683,
      "step": 235
    },
    {
      "epoch": 0.2360663936732206,
      "grad_norm": 0.5434374809265137,
      "learning_rate": 0.00019695558226570507,
      "loss": 0.5554,
      "step": 236
    },
    {
      "epoch": 0.2370666750023444,
      "grad_norm": 1.194429874420166,
      "learning_rate": 0.00019692986056661356,
      "loss": 0.5923,
      "step": 237
    },
    {
      "epoch": 0.23806695633146824,
      "grad_norm": 0.40971899032592773,
      "learning_rate": 0.00019690403235913504,
      "loss": 0.5173,
      "step": 238
    },
    {
      "epoch": 0.23906723766059204,
      "grad_norm": 0.40528661012649536,
      "learning_rate": 0.00019687809767165,
      "loss": 0.5331,
      "step": 239
    },
    {
      "epoch": 0.24006751898971584,
      "grad_norm": 0.40911996364593506,
      "learning_rate": 0.000196852056532656,
      "loss": 0.4904,
      "step": 240
    },
    {
      "epoch": 0.24106780031883968,
      "grad_norm": 0.5151987075805664,
      "learning_rate": 0.00019682590897076752,
      "loss": 0.5382,
      "step": 241
    },
    {
      "epoch": 0.24206808164796348,
      "grad_norm": 0.3909003734588623,
      "learning_rate": 0.00019679965501471608,
      "loss": 0.447,
      "step": 242
    },
    {
      "epoch": 0.2430683629770873,
      "grad_norm": 0.497062623500824,
      "learning_rate": 0.0001967732946933499,
      "loss": 0.596,
      "step": 243
    },
    {
      "epoch": 0.2440686443062111,
      "grad_norm": 0.4184584617614746,
      "learning_rate": 0.00019674682803563428,
      "loss": 0.5146,
      "step": 244
    },
    {
      "epoch": 0.24506892563533494,
      "grad_norm": 0.4521709680557251,
      "learning_rate": 0.00019672025507065131,
      "loss": 0.5296,
      "step": 245
    },
    {
      "epoch": 0.24606920696445875,
      "grad_norm": 0.44223159551620483,
      "learning_rate": 0.00019669357582759983,
      "loss": 0.5223,
      "step": 246
    },
    {
      "epoch": 0.24706948829358258,
      "grad_norm": 0.5312020778656006,
      "learning_rate": 0.00019666679033579552,
      "loss": 0.5491,
      "step": 247
    },
    {
      "epoch": 0.24806976962270638,
      "grad_norm": 0.4204310178756714,
      "learning_rate": 0.00019663989862467082,
      "loss": 0.4885,
      "step": 248
    },
    {
      "epoch": 0.2490700509518302,
      "grad_norm": 0.4755087196826935,
      "learning_rate": 0.00019661290072377482,
      "loss": 0.5295,
      "step": 249
    },
    {
      "epoch": 0.25007033228095404,
      "grad_norm": 0.8514156937599182,
      "learning_rate": 0.00019658579666277334,
      "loss": 0.5279,
      "step": 250
    },
    {
      "epoch": 0.2510706136100778,
      "grad_norm": 0.6153047680854797,
      "learning_rate": 0.0001965585864714488,
      "loss": 0.4662,
      "step": 251
    },
    {
      "epoch": 0.25207089493920165,
      "grad_norm": 0.49649399518966675,
      "learning_rate": 0.00019653127017970034,
      "loss": 0.494,
      "step": 252
    },
    {
      "epoch": 0.2530711762683255,
      "grad_norm": 0.46047142148017883,
      "learning_rate": 0.0001965038478175436,
      "loss": 0.5341,
      "step": 253
    },
    {
      "epoch": 0.25407145759744926,
      "grad_norm": 0.7829805016517639,
      "learning_rate": 0.00019647631941511082,
      "loss": 0.4661,
      "step": 254
    },
    {
      "epoch": 0.2550717389265731,
      "grad_norm": 0.5202223062515259,
      "learning_rate": 0.0001964486850026507,
      "loss": 0.5413,
      "step": 255
    },
    {
      "epoch": 0.2560720202556969,
      "grad_norm": 0.5138536095619202,
      "learning_rate": 0.00019642094461052852,
      "loss": 0.4998,
      "step": 256
    },
    {
      "epoch": 0.25707230158482075,
      "grad_norm": 0.44960975646972656,
      "learning_rate": 0.00019639309826922585,
      "loss": 0.5004,
      "step": 257
    },
    {
      "epoch": 0.2580725829139445,
      "grad_norm": 0.37193313241004944,
      "learning_rate": 0.0001963651460093409,
      "loss": 0.4918,
      "step": 258
    },
    {
      "epoch": 0.25907286424306836,
      "grad_norm": 0.41742751002311707,
      "learning_rate": 0.00019633708786158806,
      "loss": 0.5449,
      "step": 259
    },
    {
      "epoch": 0.2600731455721922,
      "grad_norm": 0.4701450765132904,
      "learning_rate": 0.00019630892385679818,
      "loss": 0.5558,
      "step": 260
    },
    {
      "epoch": 0.261073426901316,
      "grad_norm": 0.5549765229225159,
      "learning_rate": 0.00019628065402591845,
      "loss": 0.4763,
      "step": 261
    },
    {
      "epoch": 0.2620737082304398,
      "grad_norm": 0.5767666697502136,
      "learning_rate": 0.00019625227840001225,
      "loss": 0.5085,
      "step": 262
    },
    {
      "epoch": 0.2630739895595636,
      "grad_norm": 0.4518953561782837,
      "learning_rate": 0.0001962237970102593,
      "loss": 0.4957,
      "step": 263
    },
    {
      "epoch": 0.26407427088868746,
      "grad_norm": 0.47013652324676514,
      "learning_rate": 0.0001961952098879555,
      "loss": 0.4965,
      "step": 264
    },
    {
      "epoch": 0.26507455221781123,
      "grad_norm": 0.43617936968803406,
      "learning_rate": 0.00019616651706451287,
      "loss": 0.4466,
      "step": 265
    },
    {
      "epoch": 0.26607483354693506,
      "grad_norm": 0.4727296531200409,
      "learning_rate": 0.0001961377185714597,
      "loss": 0.5034,
      "step": 266
    },
    {
      "epoch": 0.2670751148760589,
      "grad_norm": 0.4447670876979828,
      "learning_rate": 0.0001961088144404403,
      "loss": 0.4418,
      "step": 267
    },
    {
      "epoch": 0.2680753962051827,
      "grad_norm": 0.4598022699356079,
      "learning_rate": 0.00019607980470321505,
      "loss": 0.5056,
      "step": 268
    },
    {
      "epoch": 0.2690756775343065,
      "grad_norm": 0.9431538581848145,
      "learning_rate": 0.00019605068939166045,
      "loss": 0.5425,
      "step": 269
    },
    {
      "epoch": 0.27007595886343033,
      "grad_norm": 0.4852324426174164,
      "learning_rate": 0.00019602146853776894,
      "loss": 0.5254,
      "step": 270
    },
    {
      "epoch": 0.27107624019255416,
      "grad_norm": 0.42556115984916687,
      "learning_rate": 0.000195992142173649,
      "loss": 0.4199,
      "step": 271
    },
    {
      "epoch": 0.272076521521678,
      "grad_norm": 1.5440813302993774,
      "learning_rate": 0.0001959627103315249,
      "loss": 0.4965,
      "step": 272
    },
    {
      "epoch": 0.27307680285080177,
      "grad_norm": 0.3926527202129364,
      "learning_rate": 0.00019593317304373705,
      "loss": 0.4596,
      "step": 273
    },
    {
      "epoch": 0.2740770841799256,
      "grad_norm": 0.4427184462547302,
      "learning_rate": 0.00019590353034274144,
      "loss": 0.5012,
      "step": 274
    },
    {
      "epoch": 0.27507736550904943,
      "grad_norm": 0.5410654544830322,
      "learning_rate": 0.00019587378226111014,
      "loss": 0.5594,
      "step": 275
    },
    {
      "epoch": 0.27607764683817326,
      "grad_norm": 0.5077141523361206,
      "learning_rate": 0.00019584392883153088,
      "loss": 0.4821,
      "step": 276
    },
    {
      "epoch": 0.27707792816729704,
      "grad_norm": 0.5456703901290894,
      "learning_rate": 0.00019581397008680717,
      "loss": 0.4654,
      "step": 277
    },
    {
      "epoch": 0.27807820949642087,
      "grad_norm": 0.5426118969917297,
      "learning_rate": 0.00019578390605985826,
      "loss": 0.4844,
      "step": 278
    },
    {
      "epoch": 0.2790784908255447,
      "grad_norm": 0.42402184009552,
      "learning_rate": 0.00019575373678371909,
      "loss": 0.4966,
      "step": 279
    },
    {
      "epoch": 0.2800787721546685,
      "grad_norm": 0.40676864981651306,
      "learning_rate": 0.00019572346229154025,
      "loss": 0.4758,
      "step": 280
    },
    {
      "epoch": 0.2810790534837923,
      "grad_norm": 0.4172145128250122,
      "learning_rate": 0.00019569308261658787,
      "loss": 0.4944,
      "step": 281
    },
    {
      "epoch": 0.28207933481291614,
      "grad_norm": 0.39111289381980896,
      "learning_rate": 0.00019566259779224378,
      "loss": 0.5315,
      "step": 282
    },
    {
      "epoch": 0.28307961614203997,
      "grad_norm": 0.36669549345970154,
      "learning_rate": 0.00019563200785200526,
      "loss": 0.4696,
      "step": 283
    },
    {
      "epoch": 0.28407989747116374,
      "grad_norm": 0.48696157336235046,
      "learning_rate": 0.00019560131282948516,
      "loss": 0.4651,
      "step": 284
    },
    {
      "epoch": 0.2850801788002876,
      "grad_norm": 0.4705601632595062,
      "learning_rate": 0.0001955705127584117,
      "loss": 0.5729,
      "step": 285
    },
    {
      "epoch": 0.2860804601294114,
      "grad_norm": 0.423532634973526,
      "learning_rate": 0.00019553960767262863,
      "loss": 0.5168,
      "step": 286
    },
    {
      "epoch": 0.28708074145853524,
      "grad_norm": 0.7851701378822327,
      "learning_rate": 0.00019550859760609503,
      "loss": 0.5141,
      "step": 287
    },
    {
      "epoch": 0.288081022787659,
      "grad_norm": 0.4529748857021332,
      "learning_rate": 0.00019547748259288536,
      "loss": 0.485,
      "step": 288
    },
    {
      "epoch": 0.28908130411678284,
      "grad_norm": 0.7077706456184387,
      "learning_rate": 0.0001954462626671894,
      "loss": 0.557,
      "step": 289
    },
    {
      "epoch": 0.2900815854459067,
      "grad_norm": 0.5855894684791565,
      "learning_rate": 0.0001954149378633122,
      "loss": 0.5297,
      "step": 290
    },
    {
      "epoch": 0.29108186677503045,
      "grad_norm": 0.7854194641113281,
      "learning_rate": 0.00019538350821567404,
      "loss": 0.4249,
      "step": 291
    },
    {
      "epoch": 0.2920821481041543,
      "grad_norm": 0.5263127684593201,
      "learning_rate": 0.00019535197375881045,
      "loss": 0.5524,
      "step": 292
    },
    {
      "epoch": 0.2930824294332781,
      "grad_norm": 0.38130998611450195,
      "learning_rate": 0.00019532033452737205,
      "loss": 0.5038,
      "step": 293
    },
    {
      "epoch": 0.29408271076240194,
      "grad_norm": 0.4063386023044586,
      "learning_rate": 0.00019528859055612468,
      "loss": 0.5091,
      "step": 294
    },
    {
      "epoch": 0.2950829920915257,
      "grad_norm": 0.4857774078845978,
      "learning_rate": 0.0001952567418799492,
      "loss": 0.5455,
      "step": 295
    },
    {
      "epoch": 0.29608327342064955,
      "grad_norm": 0.45703741908073425,
      "learning_rate": 0.00019522478853384155,
      "loss": 0.4488,
      "step": 296
    },
    {
      "epoch": 0.2970835547497734,
      "grad_norm": 0.3836863040924072,
      "learning_rate": 0.00019519273055291266,
      "loss": 0.4978,
      "step": 297
    },
    {
      "epoch": 0.2980838360788972,
      "grad_norm": 0.35562822222709656,
      "learning_rate": 0.00019516056797238846,
      "loss": 0.4739,
      "step": 298
    },
    {
      "epoch": 0.299084117408021,
      "grad_norm": 0.4030165672302246,
      "learning_rate": 0.00019512830082760987,
      "loss": 0.5039,
      "step": 299
    },
    {
      "epoch": 0.3000843987371448,
      "grad_norm": 0.5485990047454834,
      "learning_rate": 0.00019509592915403255,
      "loss": 0.5515,
      "step": 300
    },
    {
      "epoch": 0.30108468006626865,
      "grad_norm": 0.41883930563926697,
      "learning_rate": 0.00019506345298722717,
      "loss": 0.4664,
      "step": 301
    },
    {
      "epoch": 0.3020849613953925,
      "grad_norm": 0.4625292420387268,
      "learning_rate": 0.00019503087236287913,
      "loss": 0.4959,
      "step": 302
    },
    {
      "epoch": 0.30308524272451626,
      "grad_norm": 0.47342514991760254,
      "learning_rate": 0.00019499818731678873,
      "loss": 0.4354,
      "step": 303
    },
    {
      "epoch": 0.3040855240536401,
      "grad_norm": 0.4889644384384155,
      "learning_rate": 0.00019496539788487082,
      "loss": 0.5214,
      "step": 304
    },
    {
      "epoch": 0.3050858053827639,
      "grad_norm": 0.5727283954620361,
      "learning_rate": 0.0001949325041031551,
      "loss": 0.5524,
      "step": 305
    },
    {
      "epoch": 0.3060860867118877,
      "grad_norm": 0.9029666781425476,
      "learning_rate": 0.0001948995060077859,
      "loss": 0.5522,
      "step": 306
    },
    {
      "epoch": 0.3070863680410115,
      "grad_norm": 0.4093385934829712,
      "learning_rate": 0.0001948664036350221,
      "loss": 0.4834,
      "step": 307
    },
    {
      "epoch": 0.30808664937013536,
      "grad_norm": 0.46531906723976135,
      "learning_rate": 0.00019483319702123732,
      "loss": 0.4903,
      "step": 308
    },
    {
      "epoch": 0.3090869306992592,
      "grad_norm": 0.5990651249885559,
      "learning_rate": 0.00019479988620291956,
      "loss": 0.4129,
      "step": 309
    },
    {
      "epoch": 0.31008721202838296,
      "grad_norm": 0.4223703444004059,
      "learning_rate": 0.00019476647121667137,
      "loss": 0.432,
      "step": 310
    },
    {
      "epoch": 0.3110874933575068,
      "grad_norm": 0.6224504709243774,
      "learning_rate": 0.00019473295209920983,
      "loss": 0.6105,
      "step": 311
    },
    {
      "epoch": 0.3120877746866306,
      "grad_norm": 0.6995674967765808,
      "learning_rate": 0.00019469932888736632,
      "loss": 0.5236,
      "step": 312
    },
    {
      "epoch": 0.31308805601575446,
      "grad_norm": 0.393481582403183,
      "learning_rate": 0.00019466560161808674,
      "loss": 0.4544,
      "step": 313
    },
    {
      "epoch": 0.31408833734487823,
      "grad_norm": 0.9335904121398926,
      "learning_rate": 0.00019463177032843124,
      "loss": 0.4994,
      "step": 314
    },
    {
      "epoch": 0.31508861867400206,
      "grad_norm": 0.4107264280319214,
      "learning_rate": 0.00019459783505557424,
      "loss": 0.5042,
      "step": 315
    },
    {
      "epoch": 0.3160889000031259,
      "grad_norm": 0.3685978651046753,
      "learning_rate": 0.00019456379583680452,
      "loss": 0.465,
      "step": 316
    },
    {
      "epoch": 0.31708918133224967,
      "grad_norm": 0.36938750743865967,
      "learning_rate": 0.000194529652709525,
      "loss": 0.5036,
      "step": 317
    },
    {
      "epoch": 0.3180894626613735,
      "grad_norm": 0.6016782522201538,
      "learning_rate": 0.00019449540571125286,
      "loss": 0.488,
      "step": 318
    },
    {
      "epoch": 0.31908974399049733,
      "grad_norm": 0.5433719158172607,
      "learning_rate": 0.00019446105487961926,
      "loss": 0.4962,
      "step": 319
    },
    {
      "epoch": 0.32009002531962116,
      "grad_norm": 0.387410044670105,
      "learning_rate": 0.0001944266002523696,
      "loss": 0.4358,
      "step": 320
    },
    {
      "epoch": 0.32109030664874494,
      "grad_norm": 0.38905179500579834,
      "learning_rate": 0.0001943920418673633,
      "loss": 0.4784,
      "step": 321
    },
    {
      "epoch": 0.32209058797786877,
      "grad_norm": 0.5600481033325195,
      "learning_rate": 0.00019435737976257377,
      "loss": 0.493,
      "step": 322
    },
    {
      "epoch": 0.3230908693069926,
      "grad_norm": 0.4002164304256439,
      "learning_rate": 0.00019432261397608834,
      "loss": 0.4714,
      "step": 323
    },
    {
      "epoch": 0.32409115063611643,
      "grad_norm": 0.3761899769306183,
      "learning_rate": 0.00019428774454610843,
      "loss": 0.4673,
      "step": 324
    },
    {
      "epoch": 0.3250914319652402,
      "grad_norm": 0.387349396944046,
      "learning_rate": 0.00019425277151094913,
      "loss": 0.4321,
      "step": 325
    },
    {
      "epoch": 0.32609171329436404,
      "grad_norm": 0.37781885266304016,
      "learning_rate": 0.00019421769490903957,
      "loss": 0.5345,
      "step": 326
    },
    {
      "epoch": 0.32709199462348787,
      "grad_norm": 0.33928921818733215,
      "learning_rate": 0.0001941825147789225,
      "loss": 0.4169,
      "step": 327
    },
    {
      "epoch": 0.32809227595261165,
      "grad_norm": 0.4935011565685272,
      "learning_rate": 0.00019414723115925456,
      "loss": 0.4755,
      "step": 328
    },
    {
      "epoch": 0.3290925572817355,
      "grad_norm": 0.44166022539138794,
      "learning_rate": 0.0001941118440888061,
      "loss": 0.5851,
      "step": 329
    },
    {
      "epoch": 0.3300928386108593,
      "grad_norm": 0.4288978576660156,
      "learning_rate": 0.0001940763536064611,
      "loss": 0.478,
      "step": 330
    },
    {
      "epoch": 0.33109311993998314,
      "grad_norm": 0.45652419328689575,
      "learning_rate": 0.00019404075975121716,
      "loss": 0.5151,
      "step": 331
    },
    {
      "epoch": 0.3320934012691069,
      "grad_norm": 0.4268530607223511,
      "learning_rate": 0.0001940050625621855,
      "loss": 0.4809,
      "step": 332
    },
    {
      "epoch": 0.33309368259823074,
      "grad_norm": 0.40482282638549805,
      "learning_rate": 0.00019396926207859084,
      "loss": 0.4439,
      "step": 333
    },
    {
      "epoch": 0.3340939639273546,
      "grad_norm": 0.4686596691608429,
      "learning_rate": 0.0001939333583397715,
      "loss": 0.4791,
      "step": 334
    },
    {
      "epoch": 0.3350942452564784,
      "grad_norm": 0.4924260377883911,
      "learning_rate": 0.00019389735138517915,
      "loss": 0.4592,
      "step": 335
    },
    {
      "epoch": 0.3360945265856022,
      "grad_norm": 0.5177869200706482,
      "learning_rate": 0.00019386124125437895,
      "loss": 0.4488,
      "step": 336
    },
    {
      "epoch": 0.337094807914726,
      "grad_norm": 0.5712275505065918,
      "learning_rate": 0.00019382502798704935,
      "loss": 0.4791,
      "step": 337
    },
    {
      "epoch": 0.33809508924384984,
      "grad_norm": 0.5464945435523987,
      "learning_rate": 0.00019378871162298227,
      "loss": 0.5554,
      "step": 338
    },
    {
      "epoch": 0.3390953705729737,
      "grad_norm": 0.5113401412963867,
      "learning_rate": 0.00019375229220208276,
      "loss": 0.4479,
      "step": 339
    },
    {
      "epoch": 0.34009565190209745,
      "grad_norm": 0.451701819896698,
      "learning_rate": 0.00019371576976436917,
      "loss": 0.5338,
      "step": 340
    },
    {
      "epoch": 0.3410959332312213,
      "grad_norm": 0.41110074520111084,
      "learning_rate": 0.00019367914434997312,
      "loss": 0.4551,
      "step": 341
    },
    {
      "epoch": 0.3420962145603451,
      "grad_norm": 0.444566547870636,
      "learning_rate": 0.00019364241599913924,
      "loss": 0.5034,
      "step": 342
    },
    {
      "epoch": 0.3430964958894689,
      "grad_norm": 1.1947789192199707,
      "learning_rate": 0.0001936055847522254,
      "loss": 0.4569,
      "step": 343
    },
    {
      "epoch": 0.3440967772185927,
      "grad_norm": 0.5699276328086853,
      "learning_rate": 0.00019356865064970244,
      "loss": 0.4675,
      "step": 344
    },
    {
      "epoch": 0.34509705854771655,
      "grad_norm": 0.3912697732448578,
      "learning_rate": 0.0001935316137321543,
      "loss": 0.5124,
      "step": 345
    },
    {
      "epoch": 0.3460973398768404,
      "grad_norm": 0.4574461579322815,
      "learning_rate": 0.00019349447404027782,
      "loss": 0.4276,
      "step": 346
    },
    {
      "epoch": 0.34709762120596416,
      "grad_norm": 0.6651443839073181,
      "learning_rate": 0.00019345723161488283,
      "loss": 0.4731,
      "step": 347
    },
    {
      "epoch": 0.348097902535088,
      "grad_norm": 0.7808742523193359,
      "learning_rate": 0.000193419886496892,
      "loss": 0.4953,
      "step": 348
    },
    {
      "epoch": 0.3490981838642118,
      "grad_norm": 0.4347785711288452,
      "learning_rate": 0.00019338243872734086,
      "loss": 0.4121,
      "step": 349
    },
    {
      "epoch": 0.35009846519333565,
      "grad_norm": 0.5245057344436646,
      "learning_rate": 0.00019334488834737775,
      "loss": 0.4786,
      "step": 350
    },
    {
      "epoch": 0.3510987465224594,
      "grad_norm": 0.3696887195110321,
      "learning_rate": 0.00019330723539826375,
      "loss": 0.4757,
      "step": 351
    },
    {
      "epoch": 0.35209902785158326,
      "grad_norm": 0.4131191372871399,
      "learning_rate": 0.00019326947992137262,
      "loss": 0.4781,
      "step": 352
    },
    {
      "epoch": 0.3530993091807071,
      "grad_norm": 0.419505774974823,
      "learning_rate": 0.00019323162195819082,
      "loss": 0.5067,
      "step": 353
    },
    {
      "epoch": 0.35409959050983086,
      "grad_norm": 0.4195800721645355,
      "learning_rate": 0.0001931936615503174,
      "loss": 0.4244,
      "step": 354
    },
    {
      "epoch": 0.3550998718389547,
      "grad_norm": 0.384988397359848,
      "learning_rate": 0.000193155598739464,
      "loss": 0.4633,
      "step": 355
    },
    {
      "epoch": 0.3561001531680785,
      "grad_norm": 0.432335764169693,
      "learning_rate": 0.0001931174335674547,
      "loss": 0.4616,
      "step": 356
    },
    {
      "epoch": 0.35710043449720236,
      "grad_norm": 0.4696163237094879,
      "learning_rate": 0.0001930791660762262,
      "loss": 0.4931,
      "step": 357
    },
    {
      "epoch": 0.35810071582632613,
      "grad_norm": 0.44555091857910156,
      "learning_rate": 0.00019304079630782752,
      "loss": 0.4916,
      "step": 358
    },
    {
      "epoch": 0.35910099715544996,
      "grad_norm": 0.3690800368785858,
      "learning_rate": 0.0001930023243044201,
      "loss": 0.4411,
      "step": 359
    },
    {
      "epoch": 0.3601012784845738,
      "grad_norm": 0.5480444431304932,
      "learning_rate": 0.00019296375010827773,
      "loss": 0.5076,
      "step": 360
    },
    {
      "epoch": 0.3611015598136976,
      "grad_norm": 0.45511335134506226,
      "learning_rate": 0.00019292507376178643,
      "loss": 0.5281,
      "step": 361
    },
    {
      "epoch": 0.3621018411428214,
      "grad_norm": 0.42300713062286377,
      "learning_rate": 0.00019288629530744454,
      "loss": 0.532,
      "step": 362
    },
    {
      "epoch": 0.36310212247194523,
      "grad_norm": 0.44190162420272827,
      "learning_rate": 0.0001928474147878626,
      "loss": 0.4408,
      "step": 363
    },
    {
      "epoch": 0.36410240380106906,
      "grad_norm": 0.540590226650238,
      "learning_rate": 0.0001928084322457632,
      "loss": 0.4755,
      "step": 364
    },
    {
      "epoch": 0.3651026851301929,
      "grad_norm": 0.559669554233551,
      "learning_rate": 0.00019276934772398114,
      "loss": 0.4183,
      "step": 365
    },
    {
      "epoch": 0.36610296645931667,
      "grad_norm": 0.4601163864135742,
      "learning_rate": 0.00019273016126546323,
      "loss": 0.4451,
      "step": 366
    },
    {
      "epoch": 0.3671032477884405,
      "grad_norm": 0.41059058904647827,
      "learning_rate": 0.00019269087291326833,
      "loss": 0.5118,
      "step": 367
    },
    {
      "epoch": 0.36810352911756433,
      "grad_norm": 0.365362286567688,
      "learning_rate": 0.00019265148271056722,
      "loss": 0.4125,
      "step": 368
    },
    {
      "epoch": 0.3691038104466881,
      "grad_norm": 0.4561755061149597,
      "learning_rate": 0.0001926119907006426,
      "loss": 0.5015,
      "step": 369
    },
    {
      "epoch": 0.37010409177581194,
      "grad_norm": 0.37689074873924255,
      "learning_rate": 0.00019257239692688907,
      "loss": 0.4954,
      "step": 370
    },
    {
      "epoch": 0.37110437310493577,
      "grad_norm": 0.3596641421318054,
      "learning_rate": 0.00019253270143281296,
      "loss": 0.452,
      "step": 371
    },
    {
      "epoch": 0.3721046544340596,
      "grad_norm": 0.408954381942749,
      "learning_rate": 0.00019249290426203252,
      "loss": 0.4491,
      "step": 372
    },
    {
      "epoch": 0.3731049357631834,
      "grad_norm": 0.46262991428375244,
      "learning_rate": 0.0001924530054582776,
      "loss": 0.452,
      "step": 373
    },
    {
      "epoch": 0.3741052170923072,
      "grad_norm": 0.3985409438610077,
      "learning_rate": 0.0001924130050653898,
      "loss": 0.4546,
      "step": 374
    },
    {
      "epoch": 0.37510549842143104,
      "grad_norm": 0.3863021731376648,
      "learning_rate": 0.00019237290312732226,
      "loss": 0.4425,
      "step": 375
    },
    {
      "epoch": 0.37610577975055487,
      "grad_norm": 0.36261245608329773,
      "learning_rate": 0.00019233269968813984,
      "loss": 0.4499,
      "step": 376
    },
    {
      "epoch": 0.37710606107967864,
      "grad_norm": 0.34936317801475525,
      "learning_rate": 0.00019229239479201876,
      "loss": 0.4001,
      "step": 377
    },
    {
      "epoch": 0.3781063424088025,
      "grad_norm": 0.36340323090553284,
      "learning_rate": 0.0001922519884832469,
      "loss": 0.4893,
      "step": 378
    },
    {
      "epoch": 0.3791066237379263,
      "grad_norm": 0.4176476001739502,
      "learning_rate": 0.0001922114808062234,
      "loss": 0.4451,
      "step": 379
    },
    {
      "epoch": 0.3801069050670501,
      "grad_norm": 0.7352374792098999,
      "learning_rate": 0.00019217087180545893,
      "loss": 0.4793,
      "step": 380
    },
    {
      "epoch": 0.3811071863961739,
      "grad_norm": 0.36440688371658325,
      "learning_rate": 0.0001921301615255754,
      "loss": 0.4881,
      "step": 381
    },
    {
      "epoch": 0.38210746772529774,
      "grad_norm": 0.3963299095630646,
      "learning_rate": 0.0001920893500113061,
      "loss": 0.442,
      "step": 382
    },
    {
      "epoch": 0.3831077490544216,
      "grad_norm": 0.3732217848300934,
      "learning_rate": 0.00019204843730749547,
      "loss": 0.4383,
      "step": 383
    },
    {
      "epoch": 0.38410803038354535,
      "grad_norm": 0.3525499999523163,
      "learning_rate": 0.00019200742345909915,
      "loss": 0.4356,
      "step": 384
    },
    {
      "epoch": 0.3851083117126692,
      "grad_norm": 0.35598984360694885,
      "learning_rate": 0.00019196630851118398,
      "loss": 0.4441,
      "step": 385
    },
    {
      "epoch": 0.386108593041793,
      "grad_norm": 0.39109891653060913,
      "learning_rate": 0.0001919250925089278,
      "loss": 0.4396,
      "step": 386
    },
    {
      "epoch": 0.38710887437091684,
      "grad_norm": 0.5131281018257141,
      "learning_rate": 0.00019188377549761963,
      "loss": 0.4569,
      "step": 387
    },
    {
      "epoch": 0.3881091557000406,
      "grad_norm": 0.5972152948379517,
      "learning_rate": 0.00019184235752265928,
      "loss": 0.4571,
      "step": 388
    },
    {
      "epoch": 0.38910943702916445,
      "grad_norm": 0.518309473991394,
      "learning_rate": 0.00019180083862955772,
      "loss": 0.4804,
      "step": 389
    },
    {
      "epoch": 0.3901097183582883,
      "grad_norm": 0.44072574377059937,
      "learning_rate": 0.00019175921886393666,
      "loss": 0.5042,
      "step": 390
    },
    {
      "epoch": 0.39110999968741206,
      "grad_norm": 0.42578837275505066,
      "learning_rate": 0.00019171749827152869,
      "loss": 0.5209,
      "step": 391
    },
    {
      "epoch": 0.3921102810165359,
      "grad_norm": 0.5175500512123108,
      "learning_rate": 0.0001916756768981772,
      "loss": 0.4601,
      "step": 392
    },
    {
      "epoch": 0.3931105623456597,
      "grad_norm": 0.42719781398773193,
      "learning_rate": 0.00019163375478983632,
      "loss": 0.4508,
      "step": 393
    },
    {
      "epoch": 0.39411084367478355,
      "grad_norm": 0.4961738586425781,
      "learning_rate": 0.00019159173199257085,
      "loss": 0.4627,
      "step": 394
    },
    {
      "epoch": 0.3951111250039073,
      "grad_norm": 0.39553913474082947,
      "learning_rate": 0.00019154960855255628,
      "loss": 0.4836,
      "step": 395
    },
    {
      "epoch": 0.39611140633303116,
      "grad_norm": 0.36840665340423584,
      "learning_rate": 0.0001915073845160786,
      "loss": 0.4378,
      "step": 396
    },
    {
      "epoch": 0.397111687662155,
      "grad_norm": 0.5660284161567688,
      "learning_rate": 0.00019146505992953446,
      "loss": 0.3921,
      "step": 397
    },
    {
      "epoch": 0.3981119689912788,
      "grad_norm": 0.3654325306415558,
      "learning_rate": 0.00019142263483943085,
      "loss": 0.4646,
      "step": 398
    },
    {
      "epoch": 0.3991122503204026,
      "grad_norm": 0.6575559377670288,
      "learning_rate": 0.00019138010929238534,
      "loss": 0.4916,
      "step": 399
    },
    {
      "epoch": 0.4001125316495264,
      "grad_norm": 0.37716901302337646,
      "learning_rate": 0.00019133748333512575,
      "loss": 0.3914,
      "step": 400
    },
    {
      "epoch": 0.40111281297865026,
      "grad_norm": 0.40329572558403015,
      "learning_rate": 0.00019129475701449035,
      "loss": 0.4456,
      "step": 401
    },
    {
      "epoch": 0.4021130943077741,
      "grad_norm": 0.3746117055416107,
      "learning_rate": 0.0001912519303774276,
      "loss": 0.4727,
      "step": 402
    },
    {
      "epoch": 0.40311337563689786,
      "grad_norm": 0.580685019493103,
      "learning_rate": 0.0001912090034709963,
      "loss": 0.4281,
      "step": 403
    },
    {
      "epoch": 0.4041136569660217,
      "grad_norm": 0.3667939007282257,
      "learning_rate": 0.00019116597634236525,
      "loss": 0.4145,
      "step": 404
    },
    {
      "epoch": 0.4051139382951455,
      "grad_norm": 0.4624421000480652,
      "learning_rate": 0.0001911228490388136,
      "loss": 0.4459,
      "step": 405
    },
    {
      "epoch": 0.4061142196242693,
      "grad_norm": 0.422579288482666,
      "learning_rate": 0.00019107962160773035,
      "loss": 0.3903,
      "step": 406
    },
    {
      "epoch": 0.40711450095339313,
      "grad_norm": 0.40308403968811035,
      "learning_rate": 0.0001910362940966147,
      "loss": 0.4568,
      "step": 407
    },
    {
      "epoch": 0.40811478228251696,
      "grad_norm": 0.3817251920700073,
      "learning_rate": 0.00019099286655307568,
      "loss": 0.4944,
      "step": 408
    },
    {
      "epoch": 0.4091150636116408,
      "grad_norm": 0.6623880863189697,
      "learning_rate": 0.0001909493390248324,
      "loss": 0.3936,
      "step": 409
    },
    {
      "epoch": 0.41011534494076457,
      "grad_norm": 0.3836446702480316,
      "learning_rate": 0.00019090571155971366,
      "loss": 0.4518,
      "step": 410
    },
    {
      "epoch": 0.4111156262698884,
      "grad_norm": 0.7342603206634521,
      "learning_rate": 0.00019086198420565823,
      "loss": 0.4008,
      "step": 411
    },
    {
      "epoch": 0.41211590759901223,
      "grad_norm": 0.38605815172195435,
      "learning_rate": 0.00019081815701071445,
      "loss": 0.3825,
      "step": 412
    },
    {
      "epoch": 0.41311618892813606,
      "grad_norm": 0.4390144646167755,
      "learning_rate": 0.0001907742300230406,
      "loss": 0.5158,
      "step": 413
    },
    {
      "epoch": 0.41411647025725984,
      "grad_norm": 0.4054723381996155,
      "learning_rate": 0.00019073020329090444,
      "loss": 0.4426,
      "step": 414
    },
    {
      "epoch": 0.41511675158638367,
      "grad_norm": 0.37295302748680115,
      "learning_rate": 0.0001906860768626834,
      "loss": 0.4147,
      "step": 415
    },
    {
      "epoch": 0.4161170329155075,
      "grad_norm": 0.41723310947418213,
      "learning_rate": 0.00019064185078686443,
      "loss": 0.4136,
      "step": 416
    },
    {
      "epoch": 0.4171173142446313,
      "grad_norm": 0.4265085458755493,
      "learning_rate": 0.000190597525112044,
      "loss": 0.4546,
      "step": 417
    },
    {
      "epoch": 0.4181175955737551,
      "grad_norm": 0.6407977342605591,
      "learning_rate": 0.000190553099886928,
      "loss": 0.4569,
      "step": 418
    },
    {
      "epoch": 0.41911787690287894,
      "grad_norm": 0.41884028911590576,
      "learning_rate": 0.00019050857516033173,
      "loss": 0.4187,
      "step": 419
    },
    {
      "epoch": 0.42011815823200277,
      "grad_norm": 0.4397021234035492,
      "learning_rate": 0.00019046395098117983,
      "loss": 0.4551,
      "step": 420
    },
    {
      "epoch": 0.42111843956112655,
      "grad_norm": 0.39388948678970337,
      "learning_rate": 0.00019041922739850616,
      "loss": 0.4561,
      "step": 421
    },
    {
      "epoch": 0.4221187208902504,
      "grad_norm": 0.4537939429283142,
      "learning_rate": 0.00019037440446145385,
      "loss": 0.5148,
      "step": 422
    },
    {
      "epoch": 0.4231190022193742,
      "grad_norm": 0.41184279322624207,
      "learning_rate": 0.00019032948221927524,
      "loss": 0.4562,
      "step": 423
    },
    {
      "epoch": 0.42411928354849804,
      "grad_norm": 0.4077985882759094,
      "learning_rate": 0.00019028446072133175,
      "loss": 0.4862,
      "step": 424
    },
    {
      "epoch": 0.4251195648776218,
      "grad_norm": 0.3779621124267578,
      "learning_rate": 0.00019023934001709383,
      "loss": 0.4127,
      "step": 425
    },
    {
      "epoch": 0.42611984620674564,
      "grad_norm": 0.7072260975837708,
      "learning_rate": 0.00019019412015614098,
      "loss": 0.4615,
      "step": 426
    },
    {
      "epoch": 0.4271201275358695,
      "grad_norm": 0.5229703783988953,
      "learning_rate": 0.00019014880118816164,
      "loss": 0.4456,
      "step": 427
    },
    {
      "epoch": 0.4281204088649933,
      "grad_norm": 0.4435012936592102,
      "learning_rate": 0.0001901033831629532,
      "loss": 0.455,
      "step": 428
    },
    {
      "epoch": 0.4291206901941171,
      "grad_norm": 0.36504679918289185,
      "learning_rate": 0.00019005786613042185,
      "loss": 0.4972,
      "step": 429
    },
    {
      "epoch": 0.4301209715232409,
      "grad_norm": 0.38636064529418945,
      "learning_rate": 0.00019001225014058255,
      "loss": 0.449,
      "step": 430
    },
    {
      "epoch": 0.43112125285236474,
      "grad_norm": 0.4293528199195862,
      "learning_rate": 0.00018996653524355902,
      "loss": 0.4844,
      "step": 431
    },
    {
      "epoch": 0.4321215341814885,
      "grad_norm": 0.37384432554244995,
      "learning_rate": 0.00018992072148958368,
      "loss": 0.479,
      "step": 432
    },
    {
      "epoch": 0.43312181551061235,
      "grad_norm": 0.4167514145374298,
      "learning_rate": 0.00018987480892899758,
      "loss": 0.4173,
      "step": 433
    },
    {
      "epoch": 0.4341220968397362,
      "grad_norm": 0.4135938882827759,
      "learning_rate": 0.00018982879761225027,
      "loss": 0.4681,
      "step": 434
    },
    {
      "epoch": 0.43512237816886,
      "grad_norm": 0.3738775849342346,
      "learning_rate": 0.00018978268758989991,
      "loss": 0.4707,
      "step": 435
    },
    {
      "epoch": 0.4361226594979838,
      "grad_norm": 0.43221938610076904,
      "learning_rate": 0.00018973647891261307,
      "loss": 0.4261,
      "step": 436
    },
    {
      "epoch": 0.4371229408271076,
      "grad_norm": 0.7024879455566406,
      "learning_rate": 0.00018969017163116472,
      "loss": 0.4826,
      "step": 437
    },
    {
      "epoch": 0.43812322215623145,
      "grad_norm": 0.3586277961730957,
      "learning_rate": 0.0001896437657964382,
      "loss": 0.4867,
      "step": 438
    },
    {
      "epoch": 0.4391235034853553,
      "grad_norm": 0.41892313957214355,
      "learning_rate": 0.00018959726145942508,
      "loss": 0.4156,
      "step": 439
    },
    {
      "epoch": 0.44012378481447906,
      "grad_norm": 0.3496420681476593,
      "learning_rate": 0.00018955065867122528,
      "loss": 0.37,
      "step": 440
    },
    {
      "epoch": 0.4411240661436029,
      "grad_norm": 0.39177075028419495,
      "learning_rate": 0.00018950395748304678,
      "loss": 0.4554,
      "step": 441
    },
    {
      "epoch": 0.4421243474727267,
      "grad_norm": 0.3440227508544922,
      "learning_rate": 0.0001894571579462058,
      "loss": 0.4359,
      "step": 442
    },
    {
      "epoch": 0.4431246288018505,
      "grad_norm": 0.41202670335769653,
      "learning_rate": 0.00018941026011212654,
      "loss": 0.4068,
      "step": 443
    },
    {
      "epoch": 0.4441249101309743,
      "grad_norm": 0.4585283696651459,
      "learning_rate": 0.00018936326403234125,
      "loss": 0.4962,
      "step": 444
    },
    {
      "epoch": 0.44512519146009816,
      "grad_norm": 0.4074370861053467,
      "learning_rate": 0.00018931616975849006,
      "loss": 0.477,
      "step": 445
    },
    {
      "epoch": 0.446125472789222,
      "grad_norm": 0.7314764857292175,
      "learning_rate": 0.00018926897734232115,
      "loss": 0.4972,
      "step": 446
    },
    {
      "epoch": 0.44712575411834576,
      "grad_norm": 0.46429434418678284,
      "learning_rate": 0.0001892216868356904,
      "loss": 0.4283,
      "step": 447
    },
    {
      "epoch": 0.4481260354474696,
      "grad_norm": 0.5269820094108582,
      "learning_rate": 0.0001891742982905615,
      "loss": 0.4563,
      "step": 448
    },
    {
      "epoch": 0.4491263167765934,
      "grad_norm": 0.37672704458236694,
      "learning_rate": 0.00018912681175900598,
      "loss": 0.463,
      "step": 449
    },
    {
      "epoch": 0.45012659810571726,
      "grad_norm": 1.1096312999725342,
      "learning_rate": 0.00018907922729320285,
      "loss": 0.4775,
      "step": 450
    },
    {
      "epoch": 0.45112687943484103,
      "grad_norm": 0.4609079957008362,
      "learning_rate": 0.00018903154494543889,
      "loss": 0.5329,
      "step": 451
    },
    {
      "epoch": 0.45212716076396486,
      "grad_norm": 0.42541447281837463,
      "learning_rate": 0.00018898376476810834,
      "loss": 0.4258,
      "step": 452
    },
    {
      "epoch": 0.4531274420930887,
      "grad_norm": 0.42514634132385254,
      "learning_rate": 0.00018893588681371303,
      "loss": 0.4433,
      "step": 453
    },
    {
      "epoch": 0.45412772342221247,
      "grad_norm": 0.6249622106552124,
      "learning_rate": 0.00018888791113486213,
      "loss": 0.5478,
      "step": 454
    },
    {
      "epoch": 0.4551280047513363,
      "grad_norm": 0.46454930305480957,
      "learning_rate": 0.00018883983778427227,
      "loss": 0.4129,
      "step": 455
    },
    {
      "epoch": 0.45612828608046013,
      "grad_norm": 0.3539360463619232,
      "learning_rate": 0.0001887916668147673,
      "loss": 0.4428,
      "step": 456
    },
    {
      "epoch": 0.45712856740958396,
      "grad_norm": 0.3984419107437134,
      "learning_rate": 0.00018874339827927846,
      "loss": 0.4802,
      "step": 457
    },
    {
      "epoch": 0.45812884873870774,
      "grad_norm": 0.4771546721458435,
      "learning_rate": 0.00018869503223084414,
      "loss": 0.4729,
      "step": 458
    },
    {
      "epoch": 0.45912913006783157,
      "grad_norm": 0.40642663836479187,
      "learning_rate": 0.00018864656872260985,
      "loss": 0.378,
      "step": 459
    },
    {
      "epoch": 0.4601294113969554,
      "grad_norm": 0.42019134759902954,
      "learning_rate": 0.00018859800780782828,
      "loss": 0.4128,
      "step": 460
    },
    {
      "epoch": 0.46112969272607923,
      "grad_norm": 0.3868459463119507,
      "learning_rate": 0.000188549349539859,
      "loss": 0.4626,
      "step": 461
    },
    {
      "epoch": 0.462129974055203,
      "grad_norm": 0.31789880990982056,
      "learning_rate": 0.00018850059397216876,
      "loss": 0.4054,
      "step": 462
    },
    {
      "epoch": 0.46313025538432684,
      "grad_norm": 0.32979318499565125,
      "learning_rate": 0.00018845174115833099,
      "loss": 0.3991,
      "step": 463
    },
    {
      "epoch": 0.46413053671345067,
      "grad_norm": 0.33265942335128784,
      "learning_rate": 0.0001884027911520262,
      "loss": 0.411,
      "step": 464
    },
    {
      "epoch": 0.4651308180425745,
      "grad_norm": 0.3669120967388153,
      "learning_rate": 0.00018835374400704154,
      "loss": 0.3834,
      "step": 465
    },
    {
      "epoch": 0.4661310993716983,
      "grad_norm": 0.40603357553482056,
      "learning_rate": 0.00018830459977727096,
      "loss": 0.455,
      "step": 466
    },
    {
      "epoch": 0.4671313807008221,
      "grad_norm": 0.43093356490135193,
      "learning_rate": 0.0001882553585167151,
      "loss": 0.3614,
      "step": 467
    },
    {
      "epoch": 0.46813166202994594,
      "grad_norm": 0.5963067412376404,
      "learning_rate": 0.00018820602027948114,
      "loss": 0.4074,
      "step": 468
    },
    {
      "epoch": 0.4691319433590697,
      "grad_norm": 0.4733903706073761,
      "learning_rate": 0.00018815658511978298,
      "loss": 0.4063,
      "step": 469
    },
    {
      "epoch": 0.47013222468819355,
      "grad_norm": 1.338910698890686,
      "learning_rate": 0.00018810705309194083,
      "loss": 0.455,
      "step": 470
    },
    {
      "epoch": 0.4711325060173174,
      "grad_norm": 0.3329702317714691,
      "learning_rate": 0.00018805742425038145,
      "loss": 0.3776,
      "step": 471
    },
    {
      "epoch": 0.4721327873464412,
      "grad_norm": 0.362064927816391,
      "learning_rate": 0.00018800769864963802,
      "loss": 0.4351,
      "step": 472
    },
    {
      "epoch": 0.473133068675565,
      "grad_norm": 0.36146995425224304,
      "learning_rate": 0.00018795787634434994,
      "loss": 0.4226,
      "step": 473
    },
    {
      "epoch": 0.4741333500046888,
      "grad_norm": 0.34001466631889343,
      "learning_rate": 0.0001879079573892629,
      "loss": 0.401,
      "step": 474
    },
    {
      "epoch": 0.47513363133381264,
      "grad_norm": 0.3179371953010559,
      "learning_rate": 0.00018785794183922883,
      "loss": 0.4175,
      "step": 475
    },
    {
      "epoch": 0.4761339126629365,
      "grad_norm": 0.5295701622962952,
      "learning_rate": 0.00018780782974920572,
      "loss": 0.4233,
      "step": 476
    },
    {
      "epoch": 0.47713419399206025,
      "grad_norm": 0.3790874779224396,
      "learning_rate": 0.00018775762117425777,
      "loss": 0.4353,
      "step": 477
    },
    {
      "epoch": 0.4781344753211841,
      "grad_norm": 0.38352811336517334,
      "learning_rate": 0.0001877073161695551,
      "loss": 0.4574,
      "step": 478
    },
    {
      "epoch": 0.4791347566503079,
      "grad_norm": 0.4280262291431427,
      "learning_rate": 0.00018765691479037376,
      "loss": 0.4525,
      "step": 479
    },
    {
      "epoch": 0.4801350379794317,
      "grad_norm": 0.36147376894950867,
      "learning_rate": 0.00018760641709209583,
      "loss": 0.4475,
      "step": 480
    },
    {
      "epoch": 0.4811353193085555,
      "grad_norm": 0.6647056937217712,
      "learning_rate": 0.0001875558231302091,
      "loss": 0.4051,
      "step": 481
    },
    {
      "epoch": 0.48213560063767935,
      "grad_norm": 0.4935802221298218,
      "learning_rate": 0.00018750513296030718,
      "loss": 0.4151,
      "step": 482
    },
    {
      "epoch": 0.4831358819668032,
      "grad_norm": 0.5030160546302795,
      "learning_rate": 0.00018745434663808942,
      "loss": 0.5521,
      "step": 483
    },
    {
      "epoch": 0.48413616329592696,
      "grad_norm": 0.3784274160861969,
      "learning_rate": 0.0001874034642193608,
      "loss": 0.417,
      "step": 484
    },
    {
      "epoch": 0.4851364446250508,
      "grad_norm": 0.35883280634880066,
      "learning_rate": 0.0001873524857600319,
      "loss": 0.4672,
      "step": 485
    },
    {
      "epoch": 0.4861367259541746,
      "grad_norm": 0.3392554819583893,
      "learning_rate": 0.00018730141131611882,
      "loss": 0.4166,
      "step": 486
    },
    {
      "epoch": 0.48713700728329845,
      "grad_norm": 0.5107550621032715,
      "learning_rate": 0.00018725024094374315,
      "loss": 0.409,
      "step": 487
    },
    {
      "epoch": 0.4881372886124222,
      "grad_norm": 0.44673365354537964,
      "learning_rate": 0.00018719897469913184,
      "loss": 0.409,
      "step": 488
    },
    {
      "epoch": 0.48913756994154606,
      "grad_norm": 0.4512506425380707,
      "learning_rate": 0.00018714761263861728,
      "loss": 0.3925,
      "step": 489
    },
    {
      "epoch": 0.4901378512706699,
      "grad_norm": 0.42347124218940735,
      "learning_rate": 0.000187096154818637,
      "loss": 0.4185,
      "step": 490
    },
    {
      "epoch": 0.4911381325997937,
      "grad_norm": 0.5886827111244202,
      "learning_rate": 0.00018704460129573391,
      "loss": 0.3909,
      "step": 491
    },
    {
      "epoch": 0.4921384139289175,
      "grad_norm": 0.40246817469596863,
      "learning_rate": 0.00018699295212655596,
      "loss": 0.4267,
      "step": 492
    },
    {
      "epoch": 0.4931386952580413,
      "grad_norm": 0.5066978931427002,
      "learning_rate": 0.00018694120736785632,
      "loss": 0.4437,
      "step": 493
    },
    {
      "epoch": 0.49413897658716516,
      "grad_norm": 0.36570316553115845,
      "learning_rate": 0.00018688936707649304,
      "loss": 0.4221,
      "step": 494
    },
    {
      "epoch": 0.49513925791628893,
      "grad_norm": 0.4331252872943878,
      "learning_rate": 0.00018683743130942928,
      "loss": 0.4173,
      "step": 495
    },
    {
      "epoch": 0.49613953924541276,
      "grad_norm": 0.9528374671936035,
      "learning_rate": 0.00018678540012373302,
      "loss": 0.4185,
      "step": 496
    },
    {
      "epoch": 0.4971398205745366,
      "grad_norm": 0.3963034152984619,
      "learning_rate": 0.00018673327357657715,
      "loss": 0.4448,
      "step": 497
    },
    {
      "epoch": 0.4981401019036604,
      "grad_norm": 0.7620049715042114,
      "learning_rate": 0.0001866810517252393,
      "loss": 0.4721,
      "step": 498
    },
    {
      "epoch": 0.4991403832327842,
      "grad_norm": 0.5681329369544983,
      "learning_rate": 0.00018662873462710184,
      "loss": 0.4317,
      "step": 499
    },
    {
      "epoch": 0.5001406645619081,
      "grad_norm": 0.4743039608001709,
      "learning_rate": 0.0001865763223396518,
      "loss": 0.4232,
      "step": 500
    },
    {
      "epoch": 0.5011409458910319,
      "grad_norm": 0.37916484475135803,
      "learning_rate": 0.00018652381492048083,
      "loss": 0.4212,
      "step": 501
    },
    {
      "epoch": 0.5021412272201556,
      "grad_norm": 0.5244789719581604,
      "learning_rate": 0.00018647121242728506,
      "loss": 0.4376,
      "step": 502
    },
    {
      "epoch": 0.5031415085492795,
      "grad_norm": 0.5627933144569397,
      "learning_rate": 0.00018641851491786512,
      "loss": 0.4264,
      "step": 503
    },
    {
      "epoch": 0.5041417898784033,
      "grad_norm": 0.41203227639198303,
      "learning_rate": 0.00018636572245012606,
      "loss": 0.4991,
      "step": 504
    },
    {
      "epoch": 0.5051420712075271,
      "grad_norm": 0.35720953345298767,
      "learning_rate": 0.00018631283508207725,
      "loss": 0.4028,
      "step": 505
    },
    {
      "epoch": 0.506142352536651,
      "grad_norm": 0.3578648567199707,
      "learning_rate": 0.00018625985287183233,
      "loss": 0.4616,
      "step": 506
    },
    {
      "epoch": 0.5071426338657747,
      "grad_norm": 0.34693148732185364,
      "learning_rate": 0.00018620677587760916,
      "loss": 0.4524,
      "step": 507
    },
    {
      "epoch": 0.5081429151948985,
      "grad_norm": 0.6293880939483643,
      "learning_rate": 0.00018615360415772978,
      "loss": 0.5006,
      "step": 508
    },
    {
      "epoch": 0.5091431965240224,
      "grad_norm": 0.41598960757255554,
      "learning_rate": 0.00018610033777062025,
      "loss": 0.4082,
      "step": 509
    },
    {
      "epoch": 0.5101434778531462,
      "grad_norm": 0.36499714851379395,
      "learning_rate": 0.0001860469767748108,
      "loss": 0.4604,
      "step": 510
    },
    {
      "epoch": 0.5111437591822701,
      "grad_norm": 0.4774184226989746,
      "learning_rate": 0.00018599352122893539,
      "loss": 0.4367,
      "step": 511
    },
    {
      "epoch": 0.5121440405113938,
      "grad_norm": 0.3812670409679413,
      "learning_rate": 0.00018593997119173205,
      "loss": 0.4332,
      "step": 512
    },
    {
      "epoch": 0.5131443218405176,
      "grad_norm": 0.3438394367694855,
      "learning_rate": 0.00018588632672204264,
      "loss": 0.4017,
      "step": 513
    },
    {
      "epoch": 0.5141446031696415,
      "grad_norm": 0.6510987877845764,
      "learning_rate": 0.0001858325878788126,
      "loss": 0.3836,
      "step": 514
    },
    {
      "epoch": 0.5151448844987653,
      "grad_norm": 0.35568666458129883,
      "learning_rate": 0.00018577875472109134,
      "loss": 0.3948,
      "step": 515
    },
    {
      "epoch": 0.516145165827889,
      "grad_norm": 0.3859391212463379,
      "learning_rate": 0.0001857248273080317,
      "loss": 0.4098,
      "step": 516
    },
    {
      "epoch": 0.5171454471570129,
      "grad_norm": 0.3872857689857483,
      "learning_rate": 0.00018567080569889015,
      "loss": 0.4615,
      "step": 517
    },
    {
      "epoch": 0.5181457284861367,
      "grad_norm": 0.3794446587562561,
      "learning_rate": 0.00018561668995302667,
      "loss": 0.4266,
      "step": 518
    },
    {
      "epoch": 0.5191460098152605,
      "grad_norm": 0.3550468981266022,
      "learning_rate": 0.00018556248012990468,
      "loss": 0.381,
      "step": 519
    },
    {
      "epoch": 0.5201462911443844,
      "grad_norm": 0.404654860496521,
      "learning_rate": 0.000185508176289091,
      "loss": 0.4188,
      "step": 520
    },
    {
      "epoch": 0.5211465724735082,
      "grad_norm": 0.3913307189941406,
      "learning_rate": 0.00018545377849025566,
      "loss": 0.3863,
      "step": 521
    },
    {
      "epoch": 0.522146853802632,
      "grad_norm": 0.4134361445903778,
      "learning_rate": 0.0001853992867931721,
      "loss": 0.4277,
      "step": 522
    },
    {
      "epoch": 0.5231471351317558,
      "grad_norm": 0.44908756017684937,
      "learning_rate": 0.00018534470125771674,
      "loss": 0.4251,
      "step": 523
    },
    {
      "epoch": 0.5241474164608796,
      "grad_norm": 0.3776758313179016,
      "learning_rate": 0.0001852900219438693,
      "loss": 0.4222,
      "step": 524
    },
    {
      "epoch": 0.5251476977900035,
      "grad_norm": 0.4757327139377594,
      "learning_rate": 0.0001852352489117124,
      "loss": 0.3789,
      "step": 525
    },
    {
      "epoch": 0.5261479791191273,
      "grad_norm": 0.3780950605869293,
      "learning_rate": 0.00018518038222143174,
      "loss": 0.4183,
      "step": 526
    },
    {
      "epoch": 0.527148260448251,
      "grad_norm": 0.39569976925849915,
      "learning_rate": 0.00018512542193331583,
      "loss": 0.4462,
      "step": 527
    },
    {
      "epoch": 0.5281485417773749,
      "grad_norm": 0.4072045385837555,
      "learning_rate": 0.00018507036810775615,
      "loss": 0.4173,
      "step": 528
    },
    {
      "epoch": 0.5291488231064987,
      "grad_norm": 0.35623103380203247,
      "learning_rate": 0.00018501522080524688,
      "loss": 0.4636,
      "step": 529
    },
    {
      "epoch": 0.5301491044356225,
      "grad_norm": 0.400676429271698,
      "learning_rate": 0.0001849599800863849,
      "loss": 0.4407,
      "step": 530
    },
    {
      "epoch": 0.5311493857647464,
      "grad_norm": 0.34437012672424316,
      "learning_rate": 0.0001849046460118698,
      "loss": 0.4463,
      "step": 531
    },
    {
      "epoch": 0.5321496670938701,
      "grad_norm": 0.33603695034980774,
      "learning_rate": 0.0001848492186425037,
      "loss": 0.3531,
      "step": 532
    },
    {
      "epoch": 0.533149948422994,
      "grad_norm": 0.3437596559524536,
      "learning_rate": 0.0001847936980391913,
      "loss": 0.414,
      "step": 533
    },
    {
      "epoch": 0.5341502297521178,
      "grad_norm": 0.4067806899547577,
      "learning_rate": 0.00018473808426293964,
      "loss": 0.4146,
      "step": 534
    },
    {
      "epoch": 0.5351505110812416,
      "grad_norm": 0.3306344151496887,
      "learning_rate": 0.00018468237737485823,
      "loss": 0.4219,
      "step": 535
    },
    {
      "epoch": 0.5361507924103655,
      "grad_norm": 0.3787599205970764,
      "learning_rate": 0.00018462657743615888,
      "loss": 0.4006,
      "step": 536
    },
    {
      "epoch": 0.5371510737394892,
      "grad_norm": 0.43724262714385986,
      "learning_rate": 0.00018457068450815562,
      "loss": 0.4163,
      "step": 537
    },
    {
      "epoch": 0.538151355068613,
      "grad_norm": 0.36768877506256104,
      "learning_rate": 0.00018451469865226464,
      "loss": 0.435,
      "step": 538
    },
    {
      "epoch": 0.5391516363977369,
      "grad_norm": 0.7517651319503784,
      "learning_rate": 0.00018445861993000436,
      "loss": 0.4352,
      "step": 539
    },
    {
      "epoch": 0.5401519177268607,
      "grad_norm": 0.518487274646759,
      "learning_rate": 0.00018440244840299506,
      "loss": 0.4566,
      "step": 540
    },
    {
      "epoch": 0.5411521990559846,
      "grad_norm": 0.4205772876739502,
      "learning_rate": 0.0001843461841329591,
      "loss": 0.4531,
      "step": 541
    },
    {
      "epoch": 0.5421524803851083,
      "grad_norm": 0.3316408097743988,
      "learning_rate": 0.0001842898271817208,
      "loss": 0.3504,
      "step": 542
    },
    {
      "epoch": 0.5431527617142321,
      "grad_norm": 0.786952555179596,
      "learning_rate": 0.00018423337761120618,
      "loss": 0.3977,
      "step": 543
    },
    {
      "epoch": 0.544153043043356,
      "grad_norm": 0.399848610162735,
      "learning_rate": 0.00018417683548344318,
      "loss": 0.4599,
      "step": 544
    },
    {
      "epoch": 0.5451533243724798,
      "grad_norm": 0.3978089690208435,
      "learning_rate": 0.00018412020086056133,
      "loss": 0.3956,
      "step": 545
    },
    {
      "epoch": 0.5461536057016035,
      "grad_norm": 0.3669891357421875,
      "learning_rate": 0.0001840634738047918,
      "loss": 0.4151,
      "step": 546
    },
    {
      "epoch": 0.5471538870307274,
      "grad_norm": 0.4843013286590576,
      "learning_rate": 0.0001840066543784675,
      "loss": 0.408,
      "step": 547
    },
    {
      "epoch": 0.5481541683598512,
      "grad_norm": 0.39708927273750305,
      "learning_rate": 0.00018394974264402257,
      "loss": 0.4434,
      "step": 548
    },
    {
      "epoch": 0.549154449688975,
      "grad_norm": 1.6484495401382446,
      "learning_rate": 0.00018389273866399275,
      "loss": 0.4974,
      "step": 549
    },
    {
      "epoch": 0.5501547310180989,
      "grad_norm": 0.5768120884895325,
      "learning_rate": 0.00018383564250101512,
      "loss": 0.4077,
      "step": 550
    },
    {
      "epoch": 0.5511550123472226,
      "grad_norm": 0.5533931851387024,
      "learning_rate": 0.000183778454217828,
      "loss": 0.455,
      "step": 551
    },
    {
      "epoch": 0.5521552936763465,
      "grad_norm": 0.3788195550441742,
      "learning_rate": 0.0001837211738772711,
      "loss": 0.4126,
      "step": 552
    },
    {
      "epoch": 0.5531555750054703,
      "grad_norm": 0.9261886477470398,
      "learning_rate": 0.000183663801542285,
      "loss": 0.4354,
      "step": 553
    },
    {
      "epoch": 0.5541558563345941,
      "grad_norm": 0.4599369764328003,
      "learning_rate": 0.00018360633727591155,
      "loss": 0.416,
      "step": 554
    },
    {
      "epoch": 0.555156137663718,
      "grad_norm": 0.35554108023643494,
      "learning_rate": 0.00018354878114129367,
      "loss": 0.3962,
      "step": 555
    },
    {
      "epoch": 0.5561564189928417,
      "grad_norm": 0.34602364897727966,
      "learning_rate": 0.00018349113320167504,
      "loss": 0.3876,
      "step": 556
    },
    {
      "epoch": 0.5571567003219655,
      "grad_norm": 0.4358038604259491,
      "learning_rate": 0.00018343339352040042,
      "loss": 0.4835,
      "step": 557
    },
    {
      "epoch": 0.5581569816510894,
      "grad_norm": 0.4023174047470093,
      "learning_rate": 0.00018337556216091517,
      "loss": 0.4024,
      "step": 558
    },
    {
      "epoch": 0.5591572629802132,
      "grad_norm": 0.415451318025589,
      "learning_rate": 0.00018331763918676556,
      "loss": 0.4247,
      "step": 559
    },
    {
      "epoch": 0.560157544309337,
      "grad_norm": 0.4832601249217987,
      "learning_rate": 0.00018325962466159848,
      "loss": 0.4329,
      "step": 560
    },
    {
      "epoch": 0.5611578256384608,
      "grad_norm": 0.44293975830078125,
      "learning_rate": 0.00018320151864916135,
      "loss": 0.4793,
      "step": 561
    },
    {
      "epoch": 0.5621581069675846,
      "grad_norm": 0.4804152846336365,
      "learning_rate": 0.00018314332121330225,
      "loss": 0.4307,
      "step": 562
    },
    {
      "epoch": 0.5631583882967085,
      "grad_norm": 0.583884596824646,
      "learning_rate": 0.0001830850324179695,
      "loss": 0.4485,
      "step": 563
    },
    {
      "epoch": 0.5641586696258323,
      "grad_norm": 0.48487451672554016,
      "learning_rate": 0.00018302665232721208,
      "loss": 0.467,
      "step": 564
    },
    {
      "epoch": 0.565158950954956,
      "grad_norm": 0.4294680058956146,
      "learning_rate": 0.0001829681810051791,
      "loss": 0.445,
      "step": 565
    },
    {
      "epoch": 0.5661592322840799,
      "grad_norm": 0.3581942319869995,
      "learning_rate": 0.00018290961851611995,
      "loss": 0.41,
      "step": 566
    },
    {
      "epoch": 0.5671595136132037,
      "grad_norm": 0.33892402052879333,
      "learning_rate": 0.00018285096492438424,
      "loss": 0.4237,
      "step": 567
    },
    {
      "epoch": 0.5681597949423275,
      "grad_norm": 0.40708863735198975,
      "learning_rate": 0.00018279222029442163,
      "loss": 0.4055,
      "step": 568
    },
    {
      "epoch": 0.5691600762714514,
      "grad_norm": 0.47360506653785706,
      "learning_rate": 0.00018273338469078186,
      "loss": 0.4437,
      "step": 569
    },
    {
      "epoch": 0.5701603576005752,
      "grad_norm": 0.3294854760169983,
      "learning_rate": 0.00018267445817811466,
      "loss": 0.4154,
      "step": 570
    },
    {
      "epoch": 0.5711606389296989,
      "grad_norm": 0.7280967831611633,
      "learning_rate": 0.00018261544082116954,
      "loss": 0.3806,
      "step": 571
    },
    {
      "epoch": 0.5721609202588228,
      "grad_norm": 0.9555635452270508,
      "learning_rate": 0.00018255633268479595,
      "loss": 0.4265,
      "step": 572
    },
    {
      "epoch": 0.5731612015879466,
      "grad_norm": 0.7376038432121277,
      "learning_rate": 0.00018249713383394303,
      "loss": 0.4376,
      "step": 573
    },
    {
      "epoch": 0.5741614829170705,
      "grad_norm": 0.3795664608478546,
      "learning_rate": 0.0001824378443336596,
      "loss": 0.3858,
      "step": 574
    },
    {
      "epoch": 0.5751617642461943,
      "grad_norm": 0.455180287361145,
      "learning_rate": 0.00018237846424909413,
      "loss": 0.4527,
      "step": 575
    },
    {
      "epoch": 0.576162045575318,
      "grad_norm": 0.9073643684387207,
      "learning_rate": 0.00018231899364549455,
      "loss": 0.4065,
      "step": 576
    },
    {
      "epoch": 0.5771623269044419,
      "grad_norm": 0.34285011887550354,
      "learning_rate": 0.00018225943258820833,
      "loss": 0.3948,
      "step": 577
    },
    {
      "epoch": 0.5781626082335657,
      "grad_norm": 0.6750327348709106,
      "learning_rate": 0.00018219978114268227,
      "loss": 0.504,
      "step": 578
    },
    {
      "epoch": 0.5791628895626895,
      "grad_norm": 0.5154769420623779,
      "learning_rate": 0.00018214003937446253,
      "loss": 0.3736,
      "step": 579
    },
    {
      "epoch": 0.5801631708918134,
      "grad_norm": 0.36137640476226807,
      "learning_rate": 0.00018208020734919455,
      "loss": 0.4148,
      "step": 580
    },
    {
      "epoch": 0.5811634522209371,
      "grad_norm": 0.39584222435951233,
      "learning_rate": 0.00018202028513262288,
      "loss": 0.4659,
      "step": 581
    },
    {
      "epoch": 0.5821637335500609,
      "grad_norm": 1.0214509963989258,
      "learning_rate": 0.00018196027279059117,
      "loss": 0.4404,
      "step": 582
    },
    {
      "epoch": 0.5831640148791848,
      "grad_norm": 0.4204067587852478,
      "learning_rate": 0.00018190017038904215,
      "loss": 0.4273,
      "step": 583
    },
    {
      "epoch": 0.5841642962083086,
      "grad_norm": 0.41804230213165283,
      "learning_rate": 0.0001818399779940175,
      "loss": 0.4188,
      "step": 584
    },
    {
      "epoch": 0.5851645775374325,
      "grad_norm": 0.45870262384414673,
      "learning_rate": 0.0001817796956716578,
      "loss": 0.406,
      "step": 585
    },
    {
      "epoch": 0.5861648588665562,
      "grad_norm": 0.43862882256507874,
      "learning_rate": 0.00018171932348820234,
      "loss": 0.3584,
      "step": 586
    },
    {
      "epoch": 0.58716514019568,
      "grad_norm": 0.38970085978507996,
      "learning_rate": 0.0001816588615099893,
      "loss": 0.4772,
      "step": 587
    },
    {
      "epoch": 0.5881654215248039,
      "grad_norm": 0.4159144461154938,
      "learning_rate": 0.00018159830980345548,
      "loss": 0.4169,
      "step": 588
    },
    {
      "epoch": 0.5891657028539277,
      "grad_norm": 0.40597254037857056,
      "learning_rate": 0.0001815376684351362,
      "loss": 0.4401,
      "step": 589
    },
    {
      "epoch": 0.5901659841830514,
      "grad_norm": 0.38824543356895447,
      "learning_rate": 0.00018147693747166534,
      "loss": 0.4487,
      "step": 590
    },
    {
      "epoch": 0.5911662655121753,
      "grad_norm": 0.4102400541305542,
      "learning_rate": 0.00018141611697977529,
      "loss": 0.4424,
      "step": 591
    },
    {
      "epoch": 0.5921665468412991,
      "grad_norm": 0.3962213695049286,
      "learning_rate": 0.00018135520702629675,
      "loss": 0.4267,
      "step": 592
    },
    {
      "epoch": 0.5931668281704229,
      "grad_norm": 0.3365749716758728,
      "learning_rate": 0.0001812942076781588,
      "loss": 0.392,
      "step": 593
    },
    {
      "epoch": 0.5941671094995468,
      "grad_norm": 0.37013334035873413,
      "learning_rate": 0.0001812331190023886,
      "loss": 0.4019,
      "step": 594
    },
    {
      "epoch": 0.5951673908286705,
      "grad_norm": 0.4151481091976166,
      "learning_rate": 0.0001811719410661116,
      "loss": 0.4507,
      "step": 595
    },
    {
      "epoch": 0.5961676721577944,
      "grad_norm": 0.3409581184387207,
      "learning_rate": 0.00018111067393655132,
      "loss": 0.3854,
      "step": 596
    },
    {
      "epoch": 0.5971679534869182,
      "grad_norm": 0.32947930693626404,
      "learning_rate": 0.0001810493176810292,
      "loss": 0.3967,
      "step": 597
    },
    {
      "epoch": 0.598168234816042,
      "grad_norm": 0.32746782898902893,
      "learning_rate": 0.00018098787236696474,
      "loss": 0.4295,
      "step": 598
    },
    {
      "epoch": 0.5991685161451659,
      "grad_norm": 0.4366002678871155,
      "learning_rate": 0.00018092633806187513,
      "loss": 0.4067,
      "step": 599
    },
    {
      "epoch": 0.6001687974742896,
      "grad_norm": 0.3553631901741028,
      "learning_rate": 0.0001808647148333755,
      "loss": 0.3962,
      "step": 600
    },
    {
      "epoch": 0.6011690788034134,
      "grad_norm": 0.42257818579673767,
      "learning_rate": 0.00018080300274917862,
      "loss": 0.3859,
      "step": 601
    },
    {
      "epoch": 0.6021693601325373,
      "grad_norm": 0.3592822551727295,
      "learning_rate": 0.00018074120187709495,
      "loss": 0.4494,
      "step": 602
    },
    {
      "epoch": 0.6031696414616611,
      "grad_norm": 0.3296622037887573,
      "learning_rate": 0.00018067931228503246,
      "loss": 0.4183,
      "step": 603
    },
    {
      "epoch": 0.604169922790785,
      "grad_norm": 0.3634386360645294,
      "learning_rate": 0.00018061733404099655,
      "loss": 0.4555,
      "step": 604
    },
    {
      "epoch": 0.6051702041199087,
      "grad_norm": 0.35811156034469604,
      "learning_rate": 0.00018055526721309016,
      "loss": 0.4164,
      "step": 605
    },
    {
      "epoch": 0.6061704854490325,
      "grad_norm": 0.5181963443756104,
      "learning_rate": 0.0001804931118695135,
      "loss": 0.385,
      "step": 606
    },
    {
      "epoch": 0.6071707667781564,
      "grad_norm": 0.3607885241508484,
      "learning_rate": 0.00018043086807856403,
      "loss": 0.4179,
      "step": 607
    },
    {
      "epoch": 0.6081710481072802,
      "grad_norm": 0.9988733530044556,
      "learning_rate": 0.00018036853590863648,
      "loss": 0.3904,
      "step": 608
    },
    {
      "epoch": 0.609171329436404,
      "grad_norm": 0.6165673136711121,
      "learning_rate": 0.00018030611542822257,
      "loss": 0.3944,
      "step": 609
    },
    {
      "epoch": 0.6101716107655278,
      "grad_norm": 0.3638354241847992,
      "learning_rate": 0.00018024360670591114,
      "loss": 0.4462,
      "step": 610
    },
    {
      "epoch": 0.6111718920946516,
      "grad_norm": 0.5270577669143677,
      "learning_rate": 0.00018018100981038798,
      "loss": 0.3938,
      "step": 611
    },
    {
      "epoch": 0.6121721734237754,
      "grad_norm": 0.45733514428138733,
      "learning_rate": 0.00018011832481043576,
      "loss": 0.411,
      "step": 612
    },
    {
      "epoch": 0.6131724547528993,
      "grad_norm": 0.41824761033058167,
      "learning_rate": 0.00018005555177493394,
      "loss": 0.4203,
      "step": 613
    },
    {
      "epoch": 0.614172736082023,
      "grad_norm": 0.33256229758262634,
      "learning_rate": 0.00017999269077285875,
      "loss": 0.3655,
      "step": 614
    },
    {
      "epoch": 0.6151730174111469,
      "grad_norm": 0.7422583103179932,
      "learning_rate": 0.00017992974187328305,
      "loss": 0.3774,
      "step": 615
    },
    {
      "epoch": 0.6161732987402707,
      "grad_norm": 0.39932578802108765,
      "learning_rate": 0.00017986670514537627,
      "loss": 0.4369,
      "step": 616
    },
    {
      "epoch": 0.6171735800693945,
      "grad_norm": 0.34032493829727173,
      "learning_rate": 0.00017980358065840444,
      "loss": 0.3846,
      "step": 617
    },
    {
      "epoch": 0.6181738613985184,
      "grad_norm": 0.35309046506881714,
      "learning_rate": 0.0001797403684817299,
      "loss": 0.3971,
      "step": 618
    },
    {
      "epoch": 0.6191741427276422,
      "grad_norm": 0.39930373430252075,
      "learning_rate": 0.00017967706868481144,
      "loss": 0.4827,
      "step": 619
    },
    {
      "epoch": 0.6201744240567659,
      "grad_norm": 0.5394731163978577,
      "learning_rate": 0.00017961368133720407,
      "loss": 0.3876,
      "step": 620
    },
    {
      "epoch": 0.6211747053858898,
      "grad_norm": 0.31531646847724915,
      "learning_rate": 0.000179550206508559,
      "loss": 0.4414,
      "step": 621
    },
    {
      "epoch": 0.6221749867150136,
      "grad_norm": 0.34368667006492615,
      "learning_rate": 0.00017948664426862364,
      "loss": 0.4166,
      "step": 622
    },
    {
      "epoch": 0.6231752680441374,
      "grad_norm": 0.316400408744812,
      "learning_rate": 0.00017942299468724134,
      "loss": 0.4084,
      "step": 623
    },
    {
      "epoch": 0.6241755493732613,
      "grad_norm": 0.36261165142059326,
      "learning_rate": 0.0001793592578343515,
      "loss": 0.3929,
      "step": 624
    },
    {
      "epoch": 0.625175830702385,
      "grad_norm": 0.4334074556827545,
      "learning_rate": 0.0001792954337799894,
      "loss": 0.4294,
      "step": 625
    },
    {
      "epoch": 0.6261761120315089,
      "grad_norm": 0.47246891260147095,
      "learning_rate": 0.00017923152259428612,
      "loss": 0.4182,
      "step": 626
    },
    {
      "epoch": 0.6271763933606327,
      "grad_norm": 0.5358693599700928,
      "learning_rate": 0.00017916752434746856,
      "loss": 0.4227,
      "step": 627
    },
    {
      "epoch": 0.6281766746897565,
      "grad_norm": 0.3735455572605133,
      "learning_rate": 0.0001791034391098591,
      "loss": 0.4272,
      "step": 628
    },
    {
      "epoch": 0.6291769560188804,
      "grad_norm": 0.3278733193874359,
      "learning_rate": 0.00017903926695187595,
      "loss": 0.4142,
      "step": 629
    },
    {
      "epoch": 0.6301772373480041,
      "grad_norm": 0.31741592288017273,
      "learning_rate": 0.0001789750079440326,
      "loss": 0.3897,
      "step": 630
    },
    {
      "epoch": 0.6311775186771279,
      "grad_norm": 0.49502453207969666,
      "learning_rate": 0.00017891066215693817,
      "loss": 0.4131,
      "step": 631
    },
    {
      "epoch": 0.6321778000062518,
      "grad_norm": 0.3936639130115509,
      "learning_rate": 0.00017884622966129695,
      "loss": 0.4268,
      "step": 632
    },
    {
      "epoch": 0.6331780813353756,
      "grad_norm": 0.3695579767227173,
      "learning_rate": 0.00017878171052790868,
      "loss": 0.3946,
      "step": 633
    },
    {
      "epoch": 0.6341783626644993,
      "grad_norm": 0.37677764892578125,
      "learning_rate": 0.00017871710482766817,
      "loss": 0.4274,
      "step": 634
    },
    {
      "epoch": 0.6351786439936232,
      "grad_norm": 0.3215508460998535,
      "learning_rate": 0.00017865241263156546,
      "loss": 0.3649,
      "step": 635
    },
    {
      "epoch": 0.636178925322747,
      "grad_norm": 0.32095083594322205,
      "learning_rate": 0.0001785876340106855,
      "loss": 0.3968,
      "step": 636
    },
    {
      "epoch": 0.6371792066518709,
      "grad_norm": 0.3916400074958801,
      "learning_rate": 0.0001785227690362083,
      "loss": 0.4252,
      "step": 637
    },
    {
      "epoch": 0.6381794879809947,
      "grad_norm": 0.39982420206069946,
      "learning_rate": 0.00017845781777940878,
      "loss": 0.4502,
      "step": 638
    },
    {
      "epoch": 0.6391797693101184,
      "grad_norm": 0.3461998403072357,
      "learning_rate": 0.00017839278031165658,
      "loss": 0.3843,
      "step": 639
    },
    {
      "epoch": 0.6401800506392423,
      "grad_norm": 0.3941275179386139,
      "learning_rate": 0.00017832765670441612,
      "loss": 0.4267,
      "step": 640
    },
    {
      "epoch": 0.6411803319683661,
      "grad_norm": 0.38379454612731934,
      "learning_rate": 0.0001782624470292465,
      "loss": 0.4469,
      "step": 641
    },
    {
      "epoch": 0.6421806132974899,
      "grad_norm": 0.3393775224685669,
      "learning_rate": 0.0001781971513578013,
      "loss": 0.4069,
      "step": 642
    },
    {
      "epoch": 0.6431808946266138,
      "grad_norm": 0.3965872526168823,
      "learning_rate": 0.00017813176976182873,
      "loss": 0.419,
      "step": 643
    },
    {
      "epoch": 0.6441811759557375,
      "grad_norm": 0.353245347738266,
      "learning_rate": 0.00017806630231317127,
      "loss": 0.409,
      "step": 644
    },
    {
      "epoch": 0.6451814572848613,
      "grad_norm": 0.3594529628753662,
      "learning_rate": 0.00017800074908376584,
      "loss": 0.4144,
      "step": 645
    },
    {
      "epoch": 0.6461817386139852,
      "grad_norm": 0.5465638637542725,
      "learning_rate": 0.00017793511014564358,
      "loss": 0.3626,
      "step": 646
    },
    {
      "epoch": 0.647182019943109,
      "grad_norm": 0.37617313861846924,
      "learning_rate": 0.00017786938557092983,
      "loss": 0.4189,
      "step": 647
    },
    {
      "epoch": 0.6481823012722329,
      "grad_norm": 2.931878089904785,
      "learning_rate": 0.00017780357543184397,
      "loss": 0.4275,
      "step": 648
    },
    {
      "epoch": 0.6491825826013566,
      "grad_norm": 0.29411986470222473,
      "learning_rate": 0.00017773767980069945,
      "loss": 0.3582,
      "step": 649
    },
    {
      "epoch": 0.6501828639304804,
      "grad_norm": 0.2961196303367615,
      "learning_rate": 0.0001776716987499037,
      "loss": 0.3542,
      "step": 650
    },
    {
      "epoch": 0.6511831452596043,
      "grad_norm": 0.3717415928840637,
      "learning_rate": 0.0001776056323519579,
      "loss": 0.3898,
      "step": 651
    },
    {
      "epoch": 0.6521834265887281,
      "grad_norm": 0.3317027688026428,
      "learning_rate": 0.00017753948067945712,
      "loss": 0.3601,
      "step": 652
    },
    {
      "epoch": 0.6531837079178519,
      "grad_norm": 0.3202069401741028,
      "learning_rate": 0.00017747324380509006,
      "loss": 0.379,
      "step": 653
    },
    {
      "epoch": 0.6541839892469757,
      "grad_norm": 0.4527633786201477,
      "learning_rate": 0.00017740692180163908,
      "loss": 0.3632,
      "step": 654
    },
    {
      "epoch": 0.6551842705760995,
      "grad_norm": 0.36044707894325256,
      "learning_rate": 0.00017734051474198003,
      "loss": 0.436,
      "step": 655
    },
    {
      "epoch": 0.6561845519052233,
      "grad_norm": 0.49241697788238525,
      "learning_rate": 0.0001772740226990823,
      "loss": 0.3665,
      "step": 656
    },
    {
      "epoch": 0.6571848332343472,
      "grad_norm": 0.32811591029167175,
      "learning_rate": 0.00017720744574600863,
      "loss": 0.396,
      "step": 657
    },
    {
      "epoch": 0.658185114563471,
      "grad_norm": 0.38950252532958984,
      "learning_rate": 0.00017714078395591502,
      "loss": 0.4182,
      "step": 658
    },
    {
      "epoch": 0.6591853958925948,
      "grad_norm": 0.41502562165260315,
      "learning_rate": 0.00017707403740205071,
      "loss": 0.4337,
      "step": 659
    },
    {
      "epoch": 0.6601856772217186,
      "grad_norm": 0.38645392656326294,
      "learning_rate": 0.00017700720615775812,
      "loss": 0.4026,
      "step": 660
    },
    {
      "epoch": 0.6611859585508424,
      "grad_norm": 0.46210557222366333,
      "learning_rate": 0.0001769402902964727,
      "loss": 0.3655,
      "step": 661
    },
    {
      "epoch": 0.6621862398799663,
      "grad_norm": 0.39469289779663086,
      "learning_rate": 0.00017687328989172288,
      "loss": 0.3661,
      "step": 662
    },
    {
      "epoch": 0.66318652120909,
      "grad_norm": 0.35918357968330383,
      "learning_rate": 0.00017680620501712996,
      "loss": 0.3881,
      "step": 663
    },
    {
      "epoch": 0.6641868025382138,
      "grad_norm": 0.3837026357650757,
      "learning_rate": 0.00017673903574640814,
      "loss": 0.4365,
      "step": 664
    },
    {
      "epoch": 0.6651870838673377,
      "grad_norm": 0.3727085292339325,
      "learning_rate": 0.00017667178215336423,
      "loss": 0.3646,
      "step": 665
    },
    {
      "epoch": 0.6661873651964615,
      "grad_norm": 0.3274560868740082,
      "learning_rate": 0.0001766044443118978,
      "loss": 0.402,
      "step": 666
    },
    {
      "epoch": 0.6671876465255854,
      "grad_norm": 0.3634089231491089,
      "learning_rate": 0.000176537022296001,
      "loss": 0.4517,
      "step": 667
    },
    {
      "epoch": 0.6681879278547092,
      "grad_norm": 0.46388474106788635,
      "learning_rate": 0.00017646951617975837,
      "loss": 0.3643,
      "step": 668
    },
    {
      "epoch": 0.6691882091838329,
      "grad_norm": 0.5163401365280151,
      "learning_rate": 0.00017640192603734692,
      "loss": 0.4767,
      "step": 669
    },
    {
      "epoch": 0.6701884905129568,
      "grad_norm": 0.36889007687568665,
      "learning_rate": 0.00017633425194303606,
      "loss": 0.4578,
      "step": 670
    },
    {
      "epoch": 0.6711887718420806,
      "grad_norm": 0.35685673356056213,
      "learning_rate": 0.00017626649397118734,
      "loss": 0.4048,
      "step": 671
    },
    {
      "epoch": 0.6721890531712044,
      "grad_norm": 0.3536528944969177,
      "learning_rate": 0.00017619865219625452,
      "loss": 0.3955,
      "step": 672
    },
    {
      "epoch": 0.6731893345003283,
      "grad_norm": 0.407932311296463,
      "learning_rate": 0.00017613072669278343,
      "loss": 0.3873,
      "step": 673
    },
    {
      "epoch": 0.674189615829452,
      "grad_norm": 0.3567703664302826,
      "learning_rate": 0.00017606271753541192,
      "loss": 0.4208,
      "step": 674
    },
    {
      "epoch": 0.6751898971585758,
      "grad_norm": 0.40687480568885803,
      "learning_rate": 0.00017599462479886974,
      "loss": 0.4316,
      "step": 675
    },
    {
      "epoch": 0.6761901784876997,
      "grad_norm": 0.45455390214920044,
      "learning_rate": 0.00017592644855797854,
      "loss": 0.4242,
      "step": 676
    },
    {
      "epoch": 0.6771904598168235,
      "grad_norm": 0.5425058603286743,
      "learning_rate": 0.00017585818888765168,
      "loss": 0.4477,
      "step": 677
    },
    {
      "epoch": 0.6781907411459474,
      "grad_norm": 0.379422664642334,
      "learning_rate": 0.0001757898458628941,
      "loss": 0.4053,
      "step": 678
    },
    {
      "epoch": 0.6791910224750711,
      "grad_norm": 0.38145560026168823,
      "learning_rate": 0.00017572141955880252,
      "loss": 0.401,
      "step": 679
    },
    {
      "epoch": 0.6801913038041949,
      "grad_norm": 0.3887995481491089,
      "learning_rate": 0.00017565291005056504,
      "loss": 0.3914,
      "step": 680
    },
    {
      "epoch": 0.6811915851333188,
      "grad_norm": 0.3836948871612549,
      "learning_rate": 0.00017558431741346122,
      "loss": 0.3745,
      "step": 681
    },
    {
      "epoch": 0.6821918664624426,
      "grad_norm": 0.3551922142505646,
      "learning_rate": 0.00017551564172286197,
      "loss": 0.3657,
      "step": 682
    },
    {
      "epoch": 0.6831921477915663,
      "grad_norm": 0.41220593452453613,
      "learning_rate": 0.00017544688305422943,
      "loss": 0.3953,
      "step": 683
    },
    {
      "epoch": 0.6841924291206902,
      "grad_norm": 0.34753286838531494,
      "learning_rate": 0.00017537804148311695,
      "loss": 0.3988,
      "step": 684
    },
    {
      "epoch": 0.685192710449814,
      "grad_norm": 0.5912328362464905,
      "learning_rate": 0.00017530911708516902,
      "loss": 0.4302,
      "step": 685
    },
    {
      "epoch": 0.6861929917789378,
      "grad_norm": 0.31528034806251526,
      "learning_rate": 0.00017524010993612098,
      "loss": 0.4059,
      "step": 686
    },
    {
      "epoch": 0.6871932731080617,
      "grad_norm": 0.3672598898410797,
      "learning_rate": 0.00017517102011179933,
      "loss": 0.3742,
      "step": 687
    },
    {
      "epoch": 0.6881935544371854,
      "grad_norm": 0.30690911412239075,
      "learning_rate": 0.0001751018476881212,
      "loss": 0.3479,
      "step": 688
    },
    {
      "epoch": 0.6891938357663093,
      "grad_norm": 0.3117162883281708,
      "learning_rate": 0.00017503259274109464,
      "loss": 0.371,
      "step": 689
    },
    {
      "epoch": 0.6901941170954331,
      "grad_norm": 0.3757268786430359,
      "learning_rate": 0.00017496325534681825,
      "loss": 0.3933,
      "step": 690
    },
    {
      "epoch": 0.6911943984245569,
      "grad_norm": 0.4408741891384125,
      "learning_rate": 0.00017489383558148136,
      "loss": 0.4034,
      "step": 691
    },
    {
      "epoch": 0.6921946797536808,
      "grad_norm": 0.9594029188156128,
      "learning_rate": 0.00017482433352136365,
      "loss": 0.3835,
      "step": 692
    },
    {
      "epoch": 0.6931949610828045,
      "grad_norm": 0.34469136595726013,
      "learning_rate": 0.00017475474924283536,
      "loss": 0.3671,
      "step": 693
    },
    {
      "epoch": 0.6941952424119283,
      "grad_norm": 0.3134756088256836,
      "learning_rate": 0.00017468508282235704,
      "loss": 0.4062,
      "step": 694
    },
    {
      "epoch": 0.6951955237410522,
      "grad_norm": 0.3691534399986267,
      "learning_rate": 0.00017461533433647946,
      "loss": 0.4109,
      "step": 695
    },
    {
      "epoch": 0.696195805070176,
      "grad_norm": 0.3218996226787567,
      "learning_rate": 0.00017454550386184362,
      "loss": 0.3842,
      "step": 696
    },
    {
      "epoch": 0.6971960863992998,
      "grad_norm": 0.34354889392852783,
      "learning_rate": 0.00017447559147518055,
      "loss": 0.4052,
      "step": 697
    },
    {
      "epoch": 0.6981963677284236,
      "grad_norm": 0.37674522399902344,
      "learning_rate": 0.00017440559725331135,
      "loss": 0.4005,
      "step": 698
    },
    {
      "epoch": 0.6991966490575474,
      "grad_norm": 0.37338998913764954,
      "learning_rate": 0.000174335521273147,
      "loss": 0.4092,
      "step": 699
    },
    {
      "epoch": 0.7001969303866713,
      "grad_norm": 0.6372348070144653,
      "learning_rate": 0.00017426536361168834,
      "loss": 0.3503,
      "step": 700
    },
    {
      "epoch": 0.7011972117157951,
      "grad_norm": 0.4115561544895172,
      "learning_rate": 0.00017419512434602594,
      "loss": 0.4469,
      "step": 701
    },
    {
      "epoch": 0.7021974930449189,
      "grad_norm": 0.43868228793144226,
      "learning_rate": 0.00017412480355334005,
      "loss": 0.3578,
      "step": 702
    },
    {
      "epoch": 0.7031977743740427,
      "grad_norm": 0.7761533856391907,
      "learning_rate": 0.00017405440131090048,
      "loss": 0.3922,
      "step": 703
    },
    {
      "epoch": 0.7041980557031665,
      "grad_norm": 0.303922176361084,
      "learning_rate": 0.00017398391769606658,
      "loss": 0.3589,
      "step": 704
    },
    {
      "epoch": 0.7051983370322903,
      "grad_norm": 0.3312811851501465,
      "learning_rate": 0.00017391335278628712,
      "loss": 0.438,
      "step": 705
    },
    {
      "epoch": 0.7061986183614142,
      "grad_norm": 0.5369493365287781,
      "learning_rate": 0.00017384270665910014,
      "loss": 0.4066,
      "step": 706
    },
    {
      "epoch": 0.707198899690538,
      "grad_norm": 0.30470407009124756,
      "learning_rate": 0.000173771979392133,
      "loss": 0.3706,
      "step": 707
    },
    {
      "epoch": 0.7081991810196617,
      "grad_norm": 0.3040308952331543,
      "learning_rate": 0.00017370117106310214,
      "loss": 0.3873,
      "step": 708
    },
    {
      "epoch": 0.7091994623487856,
      "grad_norm": 0.3875412046909332,
      "learning_rate": 0.0001736302817498131,
      "loss": 0.4225,
      "step": 709
    },
    {
      "epoch": 0.7101997436779094,
      "grad_norm": 0.39404845237731934,
      "learning_rate": 0.00017355931153016044,
      "loss": 0.4195,
      "step": 710
    },
    {
      "epoch": 0.7112000250070333,
      "grad_norm": 0.3154905438423157,
      "learning_rate": 0.0001734882604821276,
      "loss": 0.3738,
      "step": 711
    },
    {
      "epoch": 0.712200306336157,
      "grad_norm": 0.57194983959198,
      "learning_rate": 0.0001734171286837868,
      "loss": 0.3831,
      "step": 712
    },
    {
      "epoch": 0.7132005876652808,
      "grad_norm": 0.33026668429374695,
      "learning_rate": 0.00017334591621329906,
      "loss": 0.3529,
      "step": 713
    },
    {
      "epoch": 0.7142008689944047,
      "grad_norm": 0.36765211820602417,
      "learning_rate": 0.00017327462314891402,
      "loss": 0.4175,
      "step": 714
    },
    {
      "epoch": 0.7152011503235285,
      "grad_norm": 0.44419851899147034,
      "learning_rate": 0.00017320324956896977,
      "loss": 0.3918,
      "step": 715
    },
    {
      "epoch": 0.7162014316526523,
      "grad_norm": 0.5137428045272827,
      "learning_rate": 0.00017313179555189306,
      "loss": 0.3941,
      "step": 716
    },
    {
      "epoch": 0.7172017129817762,
      "grad_norm": 0.35802266001701355,
      "learning_rate": 0.00017306026117619889,
      "loss": 0.4205,
      "step": 717
    },
    {
      "epoch": 0.7182019943108999,
      "grad_norm": 0.443865031003952,
      "learning_rate": 0.0001729886465204906,
      "loss": 0.3846,
      "step": 718
    },
    {
      "epoch": 0.7192022756400237,
      "grad_norm": 0.40550094842910767,
      "learning_rate": 0.0001729169516634598,
      "loss": 0.4528,
      "step": 719
    },
    {
      "epoch": 0.7202025569691476,
      "grad_norm": 0.31841766834259033,
      "learning_rate": 0.0001728451766838861,
      "loss": 0.3438,
      "step": 720
    },
    {
      "epoch": 0.7212028382982714,
      "grad_norm": 0.4803493022918701,
      "learning_rate": 0.00017277332166063726,
      "loss": 0.3842,
      "step": 721
    },
    {
      "epoch": 0.7222031196273953,
      "grad_norm": 0.3627019226551056,
      "learning_rate": 0.00017270138667266894,
      "loss": 0.433,
      "step": 722
    },
    {
      "epoch": 0.723203400956519,
      "grad_norm": 0.4125889241695404,
      "learning_rate": 0.00017262937179902472,
      "loss": 0.4163,
      "step": 723
    },
    {
      "epoch": 0.7242036822856428,
      "grad_norm": 0.3861655592918396,
      "learning_rate": 0.00017255727711883588,
      "loss": 0.3738,
      "step": 724
    },
    {
      "epoch": 0.7252039636147667,
      "grad_norm": 0.3277514576911926,
      "learning_rate": 0.00017248510271132144,
      "loss": 0.3935,
      "step": 725
    },
    {
      "epoch": 0.7262042449438905,
      "grad_norm": 0.5167820453643799,
      "learning_rate": 0.00017241284865578802,
      "loss": 0.3658,
      "step": 726
    },
    {
      "epoch": 0.7272045262730142,
      "grad_norm": 0.3585515320301056,
      "learning_rate": 0.00017234051503162978,
      "loss": 0.4217,
      "step": 727
    },
    {
      "epoch": 0.7282048076021381,
      "grad_norm": 0.33377212285995483,
      "learning_rate": 0.0001722681019183283,
      "loss": 0.397,
      "step": 728
    },
    {
      "epoch": 0.7292050889312619,
      "grad_norm": 0.3886008560657501,
      "learning_rate": 0.00017219560939545246,
      "loss": 0.4286,
      "step": 729
    },
    {
      "epoch": 0.7302053702603858,
      "grad_norm": 0.3317753076553345,
      "learning_rate": 0.00017212303754265843,
      "loss": 0.4312,
      "step": 730
    },
    {
      "epoch": 0.7312056515895096,
      "grad_norm": 0.3189367949962616,
      "learning_rate": 0.0001720503864396896,
      "loss": 0.4021,
      "step": 731
    },
    {
      "epoch": 0.7322059329186333,
      "grad_norm": 0.30018332600593567,
      "learning_rate": 0.00017197765616637636,
      "loss": 0.3915,
      "step": 732
    },
    {
      "epoch": 0.7332062142477572,
      "grad_norm": 0.5384621024131775,
      "learning_rate": 0.0001719048468026361,
      "loss": 0.3866,
      "step": 733
    },
    {
      "epoch": 0.734206495576881,
      "grad_norm": 0.37266600131988525,
      "learning_rate": 0.00017183195842847322,
      "loss": 0.3887,
      "step": 734
    },
    {
      "epoch": 0.7352067769060048,
      "grad_norm": 0.3440839946269989,
      "learning_rate": 0.0001717589911239788,
      "loss": 0.395,
      "step": 735
    },
    {
      "epoch": 0.7362070582351287,
      "grad_norm": 0.40594422817230225,
      "learning_rate": 0.00017168594496933074,
      "loss": 0.4307,
      "step": 736
    },
    {
      "epoch": 0.7372073395642524,
      "grad_norm": 0.34948936104774475,
      "learning_rate": 0.00017161282004479351,
      "loss": 0.4451,
      "step": 737
    },
    {
      "epoch": 0.7382076208933762,
      "grad_norm": 0.3134484589099884,
      "learning_rate": 0.0001715396164307182,
      "loss": 0.3746,
      "step": 738
    },
    {
      "epoch": 0.7392079022225001,
      "grad_norm": 0.3566936254501343,
      "learning_rate": 0.0001714663342075424,
      "loss": 0.3494,
      "step": 739
    },
    {
      "epoch": 0.7402081835516239,
      "grad_norm": 0.5428150296211243,
      "learning_rate": 0.00017139297345578994,
      "loss": 0.4158,
      "step": 740
    },
    {
      "epoch": 0.7412084648807478,
      "grad_norm": 0.35322999954223633,
      "learning_rate": 0.00017131953425607104,
      "loss": 0.4644,
      "step": 741
    },
    {
      "epoch": 0.7422087462098715,
      "grad_norm": 0.4145960211753845,
      "learning_rate": 0.00017124601668908212,
      "loss": 0.4228,
      "step": 742
    },
    {
      "epoch": 0.7432090275389953,
      "grad_norm": 0.39250168204307556,
      "learning_rate": 0.00017117242083560568,
      "loss": 0.4044,
      "step": 743
    },
    {
      "epoch": 0.7442093088681192,
      "grad_norm": 0.4388757348060608,
      "learning_rate": 0.00017109874677651024,
      "loss": 0.3974,
      "step": 744
    },
    {
      "epoch": 0.745209590197243,
      "grad_norm": 0.4119609594345093,
      "learning_rate": 0.0001710249945927503,
      "loss": 0.404,
      "step": 745
    },
    {
      "epoch": 0.7462098715263668,
      "grad_norm": 0.4308728575706482,
      "learning_rate": 0.00017095116436536612,
      "loss": 0.3965,
      "step": 746
    },
    {
      "epoch": 0.7472101528554906,
      "grad_norm": 0.46191245317459106,
      "learning_rate": 0.00017087725617548385,
      "loss": 0.397,
      "step": 747
    },
    {
      "epoch": 0.7482104341846144,
      "grad_norm": 0.29124516248703003,
      "learning_rate": 0.00017080327010431513,
      "loss": 0.3592,
      "step": 748
    },
    {
      "epoch": 0.7492107155137382,
      "grad_norm": 0.33338552713394165,
      "learning_rate": 0.00017072920623315734,
      "loss": 0.4507,
      "step": 749
    },
    {
      "epoch": 0.7502109968428621,
      "grad_norm": 0.39258667826652527,
      "learning_rate": 0.00017065506464339326,
      "loss": 0.3458,
      "step": 750
    },
    {
      "epoch": 0.7512112781719859,
      "grad_norm": 0.2836824357509613,
      "learning_rate": 0.00017058084541649106,
      "loss": 0.3406,
      "step": 751
    },
    {
      "epoch": 0.7522115595011097,
      "grad_norm": 0.31139662861824036,
      "learning_rate": 0.00017050654863400429,
      "loss": 0.3929,
      "step": 752
    },
    {
      "epoch": 0.7532118408302335,
      "grad_norm": 0.400552898645401,
      "learning_rate": 0.00017043217437757164,
      "loss": 0.3821,
      "step": 753
    },
    {
      "epoch": 0.7542121221593573,
      "grad_norm": 0.3091079592704773,
      "learning_rate": 0.00017035772272891702,
      "loss": 0.3469,
      "step": 754
    },
    {
      "epoch": 0.7552124034884812,
      "grad_norm": 0.36959725618362427,
      "learning_rate": 0.00017028319376984928,
      "loss": 0.3935,
      "step": 755
    },
    {
      "epoch": 0.756212684817605,
      "grad_norm": 0.45759984850883484,
      "learning_rate": 0.00017020858758226229,
      "loss": 0.4142,
      "step": 756
    },
    {
      "epoch": 0.7572129661467287,
      "grad_norm": 0.3765019476413727,
      "learning_rate": 0.0001701339042481347,
      "loss": 0.3456,
      "step": 757
    },
    {
      "epoch": 0.7582132474758526,
      "grad_norm": 0.6459057331085205,
      "learning_rate": 0.00017005914384953007,
      "loss": 0.4321,
      "step": 758
    },
    {
      "epoch": 0.7592135288049764,
      "grad_norm": 0.38107678294181824,
      "learning_rate": 0.00016998430646859654,
      "loss": 0.4089,
      "step": 759
    },
    {
      "epoch": 0.7602138101341002,
      "grad_norm": 0.3754514455795288,
      "learning_rate": 0.00016990939218756683,
      "loss": 0.3244,
      "step": 760
    },
    {
      "epoch": 0.761214091463224,
      "grad_norm": 0.48033371567726135,
      "learning_rate": 0.0001698344010887582,
      "loss": 0.4722,
      "step": 761
    },
    {
      "epoch": 0.7622143727923478,
      "grad_norm": 0.3488475978374481,
      "learning_rate": 0.0001697593332545723,
      "loss": 0.3762,
      "step": 762
    },
    {
      "epoch": 0.7632146541214717,
      "grad_norm": 0.39607349038124084,
      "learning_rate": 0.0001696841887674951,
      "loss": 0.3784,
      "step": 763
    },
    {
      "epoch": 0.7642149354505955,
      "grad_norm": 0.3857477307319641,
      "learning_rate": 0.00016960896771009684,
      "loss": 0.3864,
      "step": 764
    },
    {
      "epoch": 0.7652152167797193,
      "grad_norm": 0.3681580126285553,
      "learning_rate": 0.00016953367016503182,
      "loss": 0.3979,
      "step": 765
    },
    {
      "epoch": 0.7662154981088432,
      "grad_norm": 0.4560220241546631,
      "learning_rate": 0.00016945829621503838,
      "loss": 0.4099,
      "step": 766
    },
    {
      "epoch": 0.7672157794379669,
      "grad_norm": 0.4417734444141388,
      "learning_rate": 0.00016938284594293897,
      "loss": 0.4073,
      "step": 767
    },
    {
      "epoch": 0.7682160607670907,
      "grad_norm": 0.34885716438293457,
      "learning_rate": 0.00016930731943163972,
      "loss": 0.3961,
      "step": 768
    },
    {
      "epoch": 0.7692163420962146,
      "grad_norm": 0.37253254652023315,
      "learning_rate": 0.00016923171676413063,
      "loss": 0.4044,
      "step": 769
    },
    {
      "epoch": 0.7702166234253384,
      "grad_norm": 0.6925381422042847,
      "learning_rate": 0.00016915603802348535,
      "loss": 0.3547,
      "step": 770
    },
    {
      "epoch": 0.7712169047544621,
      "grad_norm": 0.3191085755825043,
      "learning_rate": 0.00016908028329286112,
      "loss": 0.3669,
      "step": 771
    },
    {
      "epoch": 0.772217186083586,
      "grad_norm": 0.3584275543689728,
      "learning_rate": 0.0001690044526554987,
      "loss": 0.3809,
      "step": 772
    },
    {
      "epoch": 0.7732174674127098,
      "grad_norm": 0.3047848045825958,
      "learning_rate": 0.00016892854619472223,
      "loss": 0.333,
      "step": 773
    },
    {
      "epoch": 0.7742177487418337,
      "grad_norm": 0.7542827129364014,
      "learning_rate": 0.00016885256399393924,
      "loss": 0.3786,
      "step": 774
    },
    {
      "epoch": 0.7752180300709575,
      "grad_norm": 0.33825355768203735,
      "learning_rate": 0.00016877650613664034,
      "loss": 0.3327,
      "step": 775
    },
    {
      "epoch": 0.7762183114000812,
      "grad_norm": 0.38772493600845337,
      "learning_rate": 0.00016870037270639942,
      "loss": 0.3837,
      "step": 776
    },
    {
      "epoch": 0.7772185927292051,
      "grad_norm": 0.3877158463001251,
      "learning_rate": 0.0001686241637868734,
      "loss": 0.399,
      "step": 777
    },
    {
      "epoch": 0.7782188740583289,
      "grad_norm": 0.8092975616455078,
      "learning_rate": 0.00016854787946180198,
      "loss": 0.4376,
      "step": 778
    },
    {
      "epoch": 0.7792191553874527,
      "grad_norm": 0.6388697028160095,
      "learning_rate": 0.00016847151981500789,
      "loss": 0.4351,
      "step": 779
    },
    {
      "epoch": 0.7802194367165766,
      "grad_norm": 0.3402675986289978,
      "learning_rate": 0.00016839508493039657,
      "loss": 0.3773,
      "step": 780
    },
    {
      "epoch": 0.7812197180457003,
      "grad_norm": 0.33726760745048523,
      "learning_rate": 0.00016831857489195618,
      "loss": 0.3578,
      "step": 781
    },
    {
      "epoch": 0.7822199993748241,
      "grad_norm": 0.3489029109477997,
      "learning_rate": 0.00016824198978375736,
      "loss": 0.4214,
      "step": 782
    },
    {
      "epoch": 0.783220280703948,
      "grad_norm": 0.3146915137767792,
      "learning_rate": 0.00016816532968995328,
      "loss": 0.3939,
      "step": 783
    },
    {
      "epoch": 0.7842205620330718,
      "grad_norm": 0.32173392176628113,
      "learning_rate": 0.0001680885946947796,
      "loss": 0.3766,
      "step": 784
    },
    {
      "epoch": 0.7852208433621957,
      "grad_norm": 0.3254373073577881,
      "learning_rate": 0.00016801178488255413,
      "loss": 0.3667,
      "step": 785
    },
    {
      "epoch": 0.7862211246913194,
      "grad_norm": 0.42575928568840027,
      "learning_rate": 0.00016793490033767698,
      "loss": 0.3913,
      "step": 786
    },
    {
      "epoch": 0.7872214060204432,
      "grad_norm": 0.3439604341983795,
      "learning_rate": 0.00016785794114463037,
      "loss": 0.3617,
      "step": 787
    },
    {
      "epoch": 0.7882216873495671,
      "grad_norm": 0.3310726583003998,
      "learning_rate": 0.00016778090738797853,
      "loss": 0.367,
      "step": 788
    },
    {
      "epoch": 0.7892219686786909,
      "grad_norm": 0.3430749177932739,
      "learning_rate": 0.00016770379915236766,
      "loss": 0.4187,
      "step": 789
    },
    {
      "epoch": 0.7902222500078147,
      "grad_norm": 0.3809003233909607,
      "learning_rate": 0.00016762661652252567,
      "loss": 0.3934,
      "step": 790
    },
    {
      "epoch": 0.7912225313369385,
      "grad_norm": 0.39951851963996887,
      "learning_rate": 0.00016754935958326244,
      "loss": 0.4307,
      "step": 791
    },
    {
      "epoch": 0.7922228126660623,
      "grad_norm": 0.34858182072639465,
      "learning_rate": 0.00016747202841946928,
      "loss": 0.3986,
      "step": 792
    },
    {
      "epoch": 0.7932230939951862,
      "grad_norm": 0.4219323992729187,
      "learning_rate": 0.00016739462311611919,
      "loss": 0.3631,
      "step": 793
    },
    {
      "epoch": 0.79422337532431,
      "grad_norm": 0.40770357847213745,
      "learning_rate": 0.00016731714375826657,
      "loss": 0.3916,
      "step": 794
    },
    {
      "epoch": 0.7952236566534338,
      "grad_norm": 0.3342955708503723,
      "learning_rate": 0.00016723959043104728,
      "loss": 0.4229,
      "step": 795
    },
    {
      "epoch": 0.7962239379825576,
      "grad_norm": 0.3682705760002136,
      "learning_rate": 0.00016716196321967832,
      "loss": 0.3864,
      "step": 796
    },
    {
      "epoch": 0.7972242193116814,
      "grad_norm": 0.38320207595825195,
      "learning_rate": 0.00016708426220945802,
      "loss": 0.4229,
      "step": 797
    },
    {
      "epoch": 0.7982245006408052,
      "grad_norm": 0.34492605924606323,
      "learning_rate": 0.00016700648748576574,
      "loss": 0.3605,
      "step": 798
    },
    {
      "epoch": 0.7992247819699291,
      "grad_norm": 0.3162643313407898,
      "learning_rate": 0.0001669286391340618,
      "loss": 0.3914,
      "step": 799
    },
    {
      "epoch": 0.8002250632990529,
      "grad_norm": 0.8038167357444763,
      "learning_rate": 0.00016685071723988748,
      "loss": 0.408,
      "step": 800
    },
    {
      "epoch": 0.8012253446281766,
      "grad_norm": 0.3795028030872345,
      "learning_rate": 0.00016677272188886483,
      "loss": 0.4213,
      "step": 801
    },
    {
      "epoch": 0.8022256259573005,
      "grad_norm": 0.3236325681209564,
      "learning_rate": 0.00016669465316669667,
      "loss": 0.3694,
      "step": 802
    },
    {
      "epoch": 0.8032259072864243,
      "grad_norm": 0.33001160621643066,
      "learning_rate": 0.00016661651115916642,
      "loss": 0.3718,
      "step": 803
    },
    {
      "epoch": 0.8042261886155482,
      "grad_norm": 0.3163105845451355,
      "learning_rate": 0.00016653829595213794,
      "loss": 0.3462,
      "step": 804
    },
    {
      "epoch": 0.805226469944672,
      "grad_norm": 0.3286716043949127,
      "learning_rate": 0.00016646000763155568,
      "loss": 0.445,
      "step": 805
    },
    {
      "epoch": 0.8062267512737957,
      "grad_norm": 0.42810407280921936,
      "learning_rate": 0.00016638164628344425,
      "loss": 0.4433,
      "step": 806
    },
    {
      "epoch": 0.8072270326029196,
      "grad_norm": 0.3572610914707184,
      "learning_rate": 0.00016630321199390867,
      "loss": 0.4008,
      "step": 807
    },
    {
      "epoch": 0.8082273139320434,
      "grad_norm": 0.3196752965450287,
      "learning_rate": 0.00016622470484913406,
      "loss": 0.3792,
      "step": 808
    },
    {
      "epoch": 0.8092275952611672,
      "grad_norm": 0.315542608499527,
      "learning_rate": 0.00016614612493538551,
      "loss": 0.3861,
      "step": 809
    },
    {
      "epoch": 0.810227876590291,
      "grad_norm": 0.3548795282840729,
      "learning_rate": 0.00016606747233900815,
      "loss": 0.3371,
      "step": 810
    },
    {
      "epoch": 0.8112281579194148,
      "grad_norm": 0.3778407573699951,
      "learning_rate": 0.00016598874714642697,
      "loss": 0.3585,
      "step": 811
    },
    {
      "epoch": 0.8122284392485386,
      "grad_norm": 0.3759431838989258,
      "learning_rate": 0.00016590994944414678,
      "loss": 0.4236,
      "step": 812
    },
    {
      "epoch": 0.8132287205776625,
      "grad_norm": 0.3228715658187866,
      "learning_rate": 0.00016583107931875192,
      "loss": 0.3585,
      "step": 813
    },
    {
      "epoch": 0.8142290019067863,
      "grad_norm": 0.4803105890750885,
      "learning_rate": 0.0001657521368569064,
      "loss": 0.3434,
      "step": 814
    },
    {
      "epoch": 0.8152292832359102,
      "grad_norm": 0.3152889609336853,
      "learning_rate": 0.0001656731221453537,
      "loss": 0.3799,
      "step": 815
    },
    {
      "epoch": 0.8162295645650339,
      "grad_norm": 0.34257394075393677,
      "learning_rate": 0.00016559403527091675,
      "loss": 0.359,
      "step": 816
    },
    {
      "epoch": 0.8172298458941577,
      "grad_norm": 0.3263014554977417,
      "learning_rate": 0.0001655148763204977,
      "loss": 0.3666,
      "step": 817
    },
    {
      "epoch": 0.8182301272232816,
      "grad_norm": 0.37621477246284485,
      "learning_rate": 0.00016543564538107797,
      "loss": 0.4339,
      "step": 818
    },
    {
      "epoch": 0.8192304085524054,
      "grad_norm": 0.3347252607345581,
      "learning_rate": 0.00016535634253971794,
      "loss": 0.3676,
      "step": 819
    },
    {
      "epoch": 0.8202306898815291,
      "grad_norm": 0.3560096025466919,
      "learning_rate": 0.00016527696788355714,
      "loss": 0.3793,
      "step": 820
    },
    {
      "epoch": 0.821230971210653,
      "grad_norm": 0.31402722001075745,
      "learning_rate": 0.00016519752149981397,
      "loss": 0.3875,
      "step": 821
    },
    {
      "epoch": 0.8222312525397768,
      "grad_norm": 0.29344919323921204,
      "learning_rate": 0.0001651180034757856,
      "loss": 0.3601,
      "step": 822
    },
    {
      "epoch": 0.8232315338689006,
      "grad_norm": 0.3649774491786957,
      "learning_rate": 0.00016503841389884798,
      "loss": 0.3715,
      "step": 823
    },
    {
      "epoch": 0.8242318151980245,
      "grad_norm": 0.35587602853775024,
      "learning_rate": 0.00016495875285645566,
      "loss": 0.317,
      "step": 824
    },
    {
      "epoch": 0.8252320965271482,
      "grad_norm": 0.31947940587997437,
      "learning_rate": 0.00016487902043614173,
      "loss": 0.3429,
      "step": 825
    },
    {
      "epoch": 0.8262323778562721,
      "grad_norm": 0.38708123564720154,
      "learning_rate": 0.0001647992167255177,
      "loss": 0.416,
      "step": 826
    },
    {
      "epoch": 0.8272326591853959,
      "grad_norm": 0.3480025827884674,
      "learning_rate": 0.0001647193418122734,
      "loss": 0.3753,
      "step": 827
    },
    {
      "epoch": 0.8282329405145197,
      "grad_norm": 0.3971632122993469,
      "learning_rate": 0.00016463939578417692,
      "loss": 0.3881,
      "step": 828
    },
    {
      "epoch": 0.8292332218436436,
      "grad_norm": 0.3156718313694,
      "learning_rate": 0.0001645593787290745,
      "loss": 0.368,
      "step": 829
    },
    {
      "epoch": 0.8302335031727673,
      "grad_norm": 0.30050358176231384,
      "learning_rate": 0.0001644792907348904,
      "loss": 0.3182,
      "step": 830
    },
    {
      "epoch": 0.8312337845018911,
      "grad_norm": 0.34513306617736816,
      "learning_rate": 0.00016439913188962685,
      "loss": 0.3906,
      "step": 831
    },
    {
      "epoch": 0.832234065831015,
      "grad_norm": 0.5657719969749451,
      "learning_rate": 0.0001643189022813639,
      "loss": 0.3495,
      "step": 832
    },
    {
      "epoch": 0.8332343471601388,
      "grad_norm": 0.38888028264045715,
      "learning_rate": 0.0001642386019982594,
      "loss": 0.395,
      "step": 833
    },
    {
      "epoch": 0.8342346284892626,
      "grad_norm": 0.37813764810562134,
      "learning_rate": 0.00016415823112854883,
      "loss": 0.4147,
      "step": 834
    },
    {
      "epoch": 0.8352349098183864,
      "grad_norm": 0.37741225957870483,
      "learning_rate": 0.00016407778976054526,
      "loss": 0.3993,
      "step": 835
    },
    {
      "epoch": 0.8362351911475102,
      "grad_norm": 0.31048983335494995,
      "learning_rate": 0.0001639972779826392,
      "loss": 0.3532,
      "step": 836
    },
    {
      "epoch": 0.8372354724766341,
      "grad_norm": 0.36776047945022583,
      "learning_rate": 0.0001639166958832985,
      "loss": 0.3684,
      "step": 837
    },
    {
      "epoch": 0.8382357538057579,
      "grad_norm": 0.3449341058731079,
      "learning_rate": 0.00016383604355106837,
      "loss": 0.3996,
      "step": 838
    },
    {
      "epoch": 0.8392360351348817,
      "grad_norm": 1.3234459161758423,
      "learning_rate": 0.00016375532107457108,
      "loss": 0.3691,
      "step": 839
    },
    {
      "epoch": 0.8402363164640055,
      "grad_norm": 0.3344302177429199,
      "learning_rate": 0.00016367452854250603,
      "loss": 0.3935,
      "step": 840
    },
    {
      "epoch": 0.8412365977931293,
      "grad_norm": 0.36884331703186035,
      "learning_rate": 0.00016359366604364972,
      "loss": 0.4174,
      "step": 841
    },
    {
      "epoch": 0.8422368791222531,
      "grad_norm": 1.48589289188385,
      "learning_rate": 0.00016351273366685526,
      "loss": 0.3874,
      "step": 842
    },
    {
      "epoch": 0.843237160451377,
      "grad_norm": 0.2866388261318207,
      "learning_rate": 0.00016343173150105278,
      "loss": 0.3513,
      "step": 843
    },
    {
      "epoch": 0.8442374417805008,
      "grad_norm": 0.3473166823387146,
      "learning_rate": 0.00016335065963524897,
      "loss": 0.3289,
      "step": 844
    },
    {
      "epoch": 0.8452377231096245,
      "grad_norm": 0.4165642559528351,
      "learning_rate": 0.0001632695181585272,
      "loss": 0.379,
      "step": 845
    },
    {
      "epoch": 0.8462380044387484,
      "grad_norm": 0.3545202314853668,
      "learning_rate": 0.00016318830716004722,
      "loss": 0.3656,
      "step": 846
    },
    {
      "epoch": 0.8472382857678722,
      "grad_norm": 0.3325307369232178,
      "learning_rate": 0.00016310702672904528,
      "loss": 0.3781,
      "step": 847
    },
    {
      "epoch": 0.8482385670969961,
      "grad_norm": 0.45360445976257324,
      "learning_rate": 0.00016302567695483382,
      "loss": 0.4268,
      "step": 848
    },
    {
      "epoch": 0.8492388484261199,
      "grad_norm": 0.3422343134880066,
      "learning_rate": 0.0001629442579268016,
      "loss": 0.3382,
      "step": 849
    },
    {
      "epoch": 0.8502391297552436,
      "grad_norm": 0.3365415632724762,
      "learning_rate": 0.00016286276973441333,
      "loss": 0.3898,
      "step": 850
    },
    {
      "epoch": 0.8512394110843675,
      "grad_norm": 0.33318525552749634,
      "learning_rate": 0.00016278121246720987,
      "loss": 0.3874,
      "step": 851
    },
    {
      "epoch": 0.8522396924134913,
      "grad_norm": 0.31284430623054504,
      "learning_rate": 0.00016269958621480788,
      "loss": 0.3834,
      "step": 852
    },
    {
      "epoch": 0.8532399737426151,
      "grad_norm": 0.84188312292099,
      "learning_rate": 0.0001626178910668998,
      "loss": 0.3753,
      "step": 853
    },
    {
      "epoch": 0.854240255071739,
      "grad_norm": 0.2953375577926636,
      "learning_rate": 0.00016253612711325386,
      "loss": 0.3312,
      "step": 854
    },
    {
      "epoch": 0.8552405364008627,
      "grad_norm": 0.3468523919582367,
      "learning_rate": 0.0001624542944437139,
      "loss": 0.3604,
      "step": 855
    },
    {
      "epoch": 0.8562408177299866,
      "grad_norm": 0.32019898295402527,
      "learning_rate": 0.00016237239314819917,
      "loss": 0.3763,
      "step": 856
    },
    {
      "epoch": 0.8572410990591104,
      "grad_norm": 0.38459116220474243,
      "learning_rate": 0.0001622904233167044,
      "loss": 0.375,
      "step": 857
    },
    {
      "epoch": 0.8582413803882342,
      "grad_norm": 0.4365327060222626,
      "learning_rate": 0.0001622083850392996,
      "loss": 0.4408,
      "step": 858
    },
    {
      "epoch": 0.859241661717358,
      "grad_norm": 0.3529103696346283,
      "learning_rate": 0.00016212627840613003,
      "loss": 0.3666,
      "step": 859
    },
    {
      "epoch": 0.8602419430464818,
      "grad_norm": 0.3275703191757202,
      "learning_rate": 0.000162044103507416,
      "loss": 0.3837,
      "step": 860
    },
    {
      "epoch": 0.8612422243756056,
      "grad_norm": 0.33619630336761475,
      "learning_rate": 0.00016196186043345288,
      "loss": 0.3703,
      "step": 861
    },
    {
      "epoch": 0.8622425057047295,
      "grad_norm": 0.3477315902709961,
      "learning_rate": 0.00016187954927461093,
      "loss": 0.3814,
      "step": 862
    },
    {
      "epoch": 0.8632427870338533,
      "grad_norm": 0.30847084522247314,
      "learning_rate": 0.00016179717012133521,
      "loss": 0.3677,
      "step": 863
    },
    {
      "epoch": 0.864243068362977,
      "grad_norm": 0.3672810196876526,
      "learning_rate": 0.00016171472306414554,
      "loss": 0.333,
      "step": 864
    },
    {
      "epoch": 0.8652433496921009,
      "grad_norm": 0.3551553785800934,
      "learning_rate": 0.00016163220819363628,
      "loss": 0.3835,
      "step": 865
    },
    {
      "epoch": 0.8662436310212247,
      "grad_norm": 0.3040880262851715,
      "learning_rate": 0.00016154962560047643,
      "loss": 0.3257,
      "step": 866
    },
    {
      "epoch": 0.8672439123503486,
      "grad_norm": 0.36442992091178894,
      "learning_rate": 0.00016146697537540924,
      "loss": 0.3474,
      "step": 867
    },
    {
      "epoch": 0.8682441936794724,
      "grad_norm": 0.35831981897354126,
      "learning_rate": 0.0001613842576092524,
      "loss": 0.4076,
      "step": 868
    },
    {
      "epoch": 0.8692444750085961,
      "grad_norm": 0.3105563521385193,
      "learning_rate": 0.00016130147239289778,
      "loss": 0.3448,
      "step": 869
    },
    {
      "epoch": 0.87024475633772,
      "grad_norm": 0.35357871651649475,
      "learning_rate": 0.00016121861981731135,
      "loss": 0.3605,
      "step": 870
    },
    {
      "epoch": 0.8712450376668438,
      "grad_norm": 0.35448822379112244,
      "learning_rate": 0.00016113569997353312,
      "loss": 0.4161,
      "step": 871
    },
    {
      "epoch": 0.8722453189959676,
      "grad_norm": 0.34788915514945984,
      "learning_rate": 0.000161052712952677,
      "loss": 0.3628,
      "step": 872
    },
    {
      "epoch": 0.8732456003250915,
      "grad_norm": 0.36797192692756653,
      "learning_rate": 0.0001609696588459307,
      "loss": 0.3704,
      "step": 873
    },
    {
      "epoch": 0.8742458816542152,
      "grad_norm": 0.30066728591918945,
      "learning_rate": 0.00016088653774455568,
      "loss": 0.3488,
      "step": 874
    },
    {
      "epoch": 0.875246162983339,
      "grad_norm": 0.3145573139190674,
      "learning_rate": 0.00016080334973988695,
      "loss": 0.372,
      "step": 875
    },
    {
      "epoch": 0.8762464443124629,
      "grad_norm": 0.41080766916275024,
      "learning_rate": 0.00016072009492333318,
      "loss": 0.4908,
      "step": 876
    },
    {
      "epoch": 0.8772467256415867,
      "grad_norm": 0.376183420419693,
      "learning_rate": 0.0001606367733863763,
      "loss": 0.3485,
      "step": 877
    },
    {
      "epoch": 0.8782470069707106,
      "grad_norm": 0.373534619808197,
      "learning_rate": 0.00016055338522057158,
      "loss": 0.4077,
      "step": 878
    },
    {
      "epoch": 0.8792472882998343,
      "grad_norm": 0.4146175682544708,
      "learning_rate": 0.00016046993051754756,
      "loss": 0.3789,
      "step": 879
    },
    {
      "epoch": 0.8802475696289581,
      "grad_norm": 0.3068220615386963,
      "learning_rate": 0.00016038640936900586,
      "loss": 0.3577,
      "step": 880
    },
    {
      "epoch": 0.881247850958082,
      "grad_norm": 0.4748701751232147,
      "learning_rate": 0.00016030282186672116,
      "loss": 0.378,
      "step": 881
    },
    {
      "epoch": 0.8822481322872058,
      "grad_norm": 0.3999297618865967,
      "learning_rate": 0.00016021916810254097,
      "loss": 0.3111,
      "step": 882
    },
    {
      "epoch": 0.8832484136163296,
      "grad_norm": 0.5979132652282715,
      "learning_rate": 0.00016013544816838565,
      "loss": 0.3342,
      "step": 883
    },
    {
      "epoch": 0.8842486949454534,
      "grad_norm": 0.34538257122039795,
      "learning_rate": 0.00016005166215624827,
      "loss": 0.3845,
      "step": 884
    },
    {
      "epoch": 0.8852489762745772,
      "grad_norm": 0.4637411832809448,
      "learning_rate": 0.0001599678101581945,
      "loss": 0.3801,
      "step": 885
    },
    {
      "epoch": 0.886249257603701,
      "grad_norm": 0.35838422179222107,
      "learning_rate": 0.00015988389226636253,
      "loss": 0.4056,
      "step": 886
    },
    {
      "epoch": 0.8872495389328249,
      "grad_norm": 0.6869409680366516,
      "learning_rate": 0.00015979990857296295,
      "loss": 0.3798,
      "step": 887
    },
    {
      "epoch": 0.8882498202619487,
      "grad_norm": 0.30768436193466187,
      "learning_rate": 0.00015971585917027862,
      "loss": 0.3356,
      "step": 888
    },
    {
      "epoch": 0.8892501015910725,
      "grad_norm": 0.33648204803466797,
      "learning_rate": 0.00015963174415066468,
      "loss": 0.3436,
      "step": 889
    },
    {
      "epoch": 0.8902503829201963,
      "grad_norm": 0.3559984266757965,
      "learning_rate": 0.0001595475636065483,
      "loss": 0.4382,
      "step": 890
    },
    {
      "epoch": 0.8912506642493201,
      "grad_norm": 0.36695605516433716,
      "learning_rate": 0.00015946331763042867,
      "loss": 0.3494,
      "step": 891
    },
    {
      "epoch": 0.892250945578444,
      "grad_norm": 0.4764396548271179,
      "learning_rate": 0.00015937900631487686,
      "loss": 0.3581,
      "step": 892
    },
    {
      "epoch": 0.8932512269075678,
      "grad_norm": 0.6888701319694519,
      "learning_rate": 0.00015929462975253585,
      "loss": 0.3178,
      "step": 893
    },
    {
      "epoch": 0.8942515082366915,
      "grad_norm": 0.6159264445304871,
      "learning_rate": 0.00015921018803612014,
      "loss": 0.396,
      "step": 894
    },
    {
      "epoch": 0.8952517895658154,
      "grad_norm": 0.31623807549476624,
      "learning_rate": 0.0001591256812584159,
      "loss": 0.2978,
      "step": 895
    },
    {
      "epoch": 0.8962520708949392,
      "grad_norm": 0.35506826639175415,
      "learning_rate": 0.00015904110951228082,
      "loss": 0.3606,
      "step": 896
    },
    {
      "epoch": 0.897252352224063,
      "grad_norm": 0.34114235639572144,
      "learning_rate": 0.00015895647289064396,
      "loss": 0.3838,
      "step": 897
    },
    {
      "epoch": 0.8982526335531869,
      "grad_norm": 0.3276808559894562,
      "learning_rate": 0.00015887177148650564,
      "loss": 0.3857,
      "step": 898
    },
    {
      "epoch": 0.8992529148823106,
      "grad_norm": 0.29563194513320923,
      "learning_rate": 0.0001587870053929374,
      "loss": 0.3158,
      "step": 899
    },
    {
      "epoch": 0.9002531962114345,
      "grad_norm": 0.3460831046104431,
      "learning_rate": 0.00015870217470308188,
      "loss": 0.3534,
      "step": 900
    },
    {
      "epoch": 0.9012534775405583,
      "grad_norm": 0.504124641418457,
      "learning_rate": 0.0001586172795101526,
      "loss": 0.344,
      "step": 901
    },
    {
      "epoch": 0.9022537588696821,
      "grad_norm": 0.33151736855506897,
      "learning_rate": 0.00015853231990743406,
      "loss": 0.3633,
      "step": 902
    },
    {
      "epoch": 0.903254040198806,
      "grad_norm": 0.3092218339443207,
      "learning_rate": 0.0001584472959882815,
      "loss": 0.3628,
      "step": 903
    },
    {
      "epoch": 0.9042543215279297,
      "grad_norm": 0.3464767038822174,
      "learning_rate": 0.00015836220784612085,
      "loss": 0.389,
      "step": 904
    },
    {
      "epoch": 0.9052546028570535,
      "grad_norm": 0.3286440670490265,
      "learning_rate": 0.00015827705557444852,
      "loss": 0.3876,
      "step": 905
    },
    {
      "epoch": 0.9062548841861774,
      "grad_norm": 0.366463303565979,
      "learning_rate": 0.00015819183926683153,
      "loss": 0.3503,
      "step": 906
    },
    {
      "epoch": 0.9072551655153012,
      "grad_norm": 0.3196199834346771,
      "learning_rate": 0.00015810655901690715,
      "loss": 0.3661,
      "step": 907
    },
    {
      "epoch": 0.9082554468444249,
      "grad_norm": 0.3566863536834717,
      "learning_rate": 0.00015802121491838297,
      "loss": 0.3606,
      "step": 908
    },
    {
      "epoch": 0.9092557281735488,
      "grad_norm": 0.39771768450737,
      "learning_rate": 0.0001579358070650367,
      "loss": 0.3594,
      "step": 909
    },
    {
      "epoch": 0.9102560095026726,
      "grad_norm": 0.7240241765975952,
      "learning_rate": 0.00015785033555071616,
      "loss": 0.376,
      "step": 910
    },
    {
      "epoch": 0.9112562908317965,
      "grad_norm": 0.2882246673107147,
      "learning_rate": 0.00015776480046933905,
      "loss": 0.341,
      "step": 911
    },
    {
      "epoch": 0.9122565721609203,
      "grad_norm": 0.3204120099544525,
      "learning_rate": 0.000157679201914893,
      "loss": 0.3207,
      "step": 912
    },
    {
      "epoch": 0.913256853490044,
      "grad_norm": 0.3301410377025604,
      "learning_rate": 0.00015759353998143528,
      "loss": 0.396,
      "step": 913
    },
    {
      "epoch": 0.9142571348191679,
      "grad_norm": 0.32384902238845825,
      "learning_rate": 0.00015750781476309288,
      "loss": 0.4229,
      "step": 914
    },
    {
      "epoch": 0.9152574161482917,
      "grad_norm": 0.5250555276870728,
      "learning_rate": 0.00015742202635406235,
      "loss": 0.3598,
      "step": 915
    },
    {
      "epoch": 0.9162576974774155,
      "grad_norm": 0.3382413685321808,
      "learning_rate": 0.00015733617484860963,
      "loss": 0.3836,
      "step": 916
    },
    {
      "epoch": 0.9172579788065394,
      "grad_norm": 0.3074124753475189,
      "learning_rate": 0.00015725026034106996,
      "loss": 0.36,
      "step": 917
    },
    {
      "epoch": 0.9182582601356631,
      "grad_norm": 0.3361177146434784,
      "learning_rate": 0.00015716428292584787,
      "loss": 0.3657,
      "step": 918
    },
    {
      "epoch": 0.919258541464787,
      "grad_norm": 0.3650394380092621,
      "learning_rate": 0.00015707824269741702,
      "loss": 0.3427,
      "step": 919
    },
    {
      "epoch": 0.9202588227939108,
      "grad_norm": 0.3097604215145111,
      "learning_rate": 0.00015699213975031996,
      "loss": 0.3481,
      "step": 920
    },
    {
      "epoch": 0.9212591041230346,
      "grad_norm": 0.3795812427997589,
      "learning_rate": 0.0001569059741791684,
      "loss": 0.4088,
      "step": 921
    },
    {
      "epoch": 0.9222593854521585,
      "grad_norm": 0.31244784593582153,
      "learning_rate": 0.0001568197460786426,
      "loss": 0.3568,
      "step": 922
    },
    {
      "epoch": 0.9232596667812822,
      "grad_norm": 0.30478861927986145,
      "learning_rate": 0.0001567334555434917,
      "loss": 0.3382,
      "step": 923
    },
    {
      "epoch": 0.924259948110406,
      "grad_norm": 0.37436509132385254,
      "learning_rate": 0.0001566471026685334,
      "loss": 0.3972,
      "step": 924
    },
    {
      "epoch": 0.9252602294395299,
      "grad_norm": 0.47988948225975037,
      "learning_rate": 0.00015656068754865387,
      "loss": 0.3806,
      "step": 925
    },
    {
      "epoch": 0.9262605107686537,
      "grad_norm": 0.4235193133354187,
      "learning_rate": 0.00015647421027880772,
      "loss": 0.3587,
      "step": 926
    },
    {
      "epoch": 0.9272607920977775,
      "grad_norm": 0.3732706904411316,
      "learning_rate": 0.0001563876709540178,
      "loss": 0.3739,
      "step": 927
    },
    {
      "epoch": 0.9282610734269013,
      "grad_norm": 0.3852796256542206,
      "learning_rate": 0.0001563010696693752,
      "loss": 0.3273,
      "step": 928
    },
    {
      "epoch": 0.9292613547560251,
      "grad_norm": 0.5863430500030518,
      "learning_rate": 0.00015621440652003907,
      "loss": 0.3634,
      "step": 929
    },
    {
      "epoch": 0.930261636085149,
      "grad_norm": 0.3562791347503662,
      "learning_rate": 0.00015612768160123652,
      "loss": 0.399,
      "step": 930
    },
    {
      "epoch": 0.9312619174142728,
      "grad_norm": 0.32722219824790955,
      "learning_rate": 0.00015604089500826257,
      "loss": 0.3635,
      "step": 931
    },
    {
      "epoch": 0.9322621987433966,
      "grad_norm": 0.3670520484447479,
      "learning_rate": 0.00015595404683648,
      "loss": 0.3754,
      "step": 932
    },
    {
      "epoch": 0.9332624800725204,
      "grad_norm": 0.9929372072219849,
      "learning_rate": 0.00015586713718131922,
      "loss": 0.3406,
      "step": 933
    },
    {
      "epoch": 0.9342627614016442,
      "grad_norm": 0.32644668221473694,
      "learning_rate": 0.0001557801661382782,
      "loss": 0.4139,
      "step": 934
    },
    {
      "epoch": 0.935263042730768,
      "grad_norm": 0.4435413181781769,
      "learning_rate": 0.00015569313380292248,
      "loss": 0.4399,
      "step": 935
    },
    {
      "epoch": 0.9362633240598919,
      "grad_norm": 0.40720775723457336,
      "learning_rate": 0.00015560604027088477,
      "loss": 0.4183,
      "step": 936
    },
    {
      "epoch": 0.9372636053890157,
      "grad_norm": 0.34059423208236694,
      "learning_rate": 0.00015551888563786515,
      "loss": 0.3913,
      "step": 937
    },
    {
      "epoch": 0.9382638867181394,
      "grad_norm": 0.29352614283561707,
      "learning_rate": 0.00015543166999963076,
      "loss": 0.3601,
      "step": 938
    },
    {
      "epoch": 0.9392641680472633,
      "grad_norm": 0.43342575430870056,
      "learning_rate": 0.0001553443934520159,
      "loss": 0.3962,
      "step": 939
    },
    {
      "epoch": 0.9402644493763871,
      "grad_norm": 0.4259299635887146,
      "learning_rate": 0.00015525705609092157,
      "loss": 0.403,
      "step": 940
    },
    {
      "epoch": 0.941264730705511,
      "grad_norm": 0.3217255771160126,
      "learning_rate": 0.00015516965801231586,
      "loss": 0.3154,
      "step": 941
    },
    {
      "epoch": 0.9422650120346348,
      "grad_norm": 0.36986494064331055,
      "learning_rate": 0.0001550821993122334,
      "loss": 0.4052,
      "step": 942
    },
    {
      "epoch": 0.9432652933637585,
      "grad_norm": 0.2910899817943573,
      "learning_rate": 0.0001549946800867755,
      "loss": 0.3634,
      "step": 943
    },
    {
      "epoch": 0.9442655746928824,
      "grad_norm": 0.3085552752017975,
      "learning_rate": 0.00015490710043210997,
      "loss": 0.3772,
      "step": 944
    },
    {
      "epoch": 0.9452658560220062,
      "grad_norm": 0.6676504015922546,
      "learning_rate": 0.00015481946044447099,
      "loss": 0.3947,
      "step": 945
    },
    {
      "epoch": 0.94626613735113,
      "grad_norm": 0.30866727232933044,
      "learning_rate": 0.00015473176022015906,
      "loss": 0.3405,
      "step": 946
    },
    {
      "epoch": 0.9472664186802539,
      "grad_norm": 1.1128013134002686,
      "learning_rate": 0.0001546439998555409,
      "loss": 0.3344,
      "step": 947
    },
    {
      "epoch": 0.9482667000093776,
      "grad_norm": 0.3300032913684845,
      "learning_rate": 0.0001545561794470492,
      "loss": 0.3075,
      "step": 948
    },
    {
      "epoch": 0.9492669813385014,
      "grad_norm": 0.3318561911582947,
      "learning_rate": 0.00015446829909118275,
      "loss": 0.354,
      "step": 949
    },
    {
      "epoch": 0.9502672626676253,
      "grad_norm": 0.2914285361766815,
      "learning_rate": 0.00015438035888450623,
      "loss": 0.332,
      "step": 950
    },
    {
      "epoch": 0.9512675439967491,
      "grad_norm": 0.3457191586494446,
      "learning_rate": 0.00015429235892364994,
      "loss": 0.3634,
      "step": 951
    },
    {
      "epoch": 0.952267825325873,
      "grad_norm": 0.37430402636528015,
      "learning_rate": 0.00015420429930530996,
      "loss": 0.3838,
      "step": 952
    },
    {
      "epoch": 0.9532681066549967,
      "grad_norm": 0.3771854639053345,
      "learning_rate": 0.00015411618012624786,
      "loss": 0.371,
      "step": 953
    },
    {
      "epoch": 0.9542683879841205,
      "grad_norm": 0.30570414662361145,
      "learning_rate": 0.00015402800148329071,
      "loss": 0.4134,
      "step": 954
    },
    {
      "epoch": 0.9552686693132444,
      "grad_norm": 0.4078763425350189,
      "learning_rate": 0.00015393976347333088,
      "loss": 0.3179,
      "step": 955
    },
    {
      "epoch": 0.9562689506423682,
      "grad_norm": 0.33647480607032776,
      "learning_rate": 0.00015385146619332596,
      "loss": 0.3572,
      "step": 956
    },
    {
      "epoch": 0.9572692319714919,
      "grad_norm": 0.34456923604011536,
      "learning_rate": 0.00015376310974029873,
      "loss": 0.3438,
      "step": 957
    },
    {
      "epoch": 0.9582695133006158,
      "grad_norm": 0.3808373808860779,
      "learning_rate": 0.00015367469421133695,
      "loss": 0.3592,
      "step": 958
    },
    {
      "epoch": 0.9592697946297396,
      "grad_norm": 0.32563215494155884,
      "learning_rate": 0.00015358621970359325,
      "loss": 0.3531,
      "step": 959
    },
    {
      "epoch": 0.9602700759588634,
      "grad_norm": 0.31296202540397644,
      "learning_rate": 0.00015349768631428519,
      "loss": 0.3645,
      "step": 960
    },
    {
      "epoch": 0.9612703572879873,
      "grad_norm": 0.4087706506252289,
      "learning_rate": 0.00015340909414069488,
      "loss": 0.3854,
      "step": 961
    },
    {
      "epoch": 0.962270638617111,
      "grad_norm": 0.37943634390830994,
      "learning_rate": 0.00015332044328016914,
      "loss": 0.349,
      "step": 962
    },
    {
      "epoch": 0.9632709199462349,
      "grad_norm": 0.9838475584983826,
      "learning_rate": 0.0001532317338301192,
      "loss": 0.3596,
      "step": 963
    },
    {
      "epoch": 0.9642712012753587,
      "grad_norm": 0.3703637421131134,
      "learning_rate": 0.00015314296588802076,
      "loss": 0.4281,
      "step": 964
    },
    {
      "epoch": 0.9652714826044825,
      "grad_norm": 0.33770328760147095,
      "learning_rate": 0.00015305413955141365,
      "loss": 0.3975,
      "step": 965
    },
    {
      "epoch": 0.9662717639336064,
      "grad_norm": 0.28159573674201965,
      "learning_rate": 0.00015296525491790205,
      "loss": 0.3268,
      "step": 966
    },
    {
      "epoch": 0.9672720452627301,
      "grad_norm": 0.7794631123542786,
      "learning_rate": 0.00015287631208515406,
      "loss": 0.3776,
      "step": 967
    },
    {
      "epoch": 0.9682723265918539,
      "grad_norm": 0.31693628430366516,
      "learning_rate": 0.00015278731115090171,
      "loss": 0.3253,
      "step": 968
    },
    {
      "epoch": 0.9692726079209778,
      "grad_norm": 0.3297598361968994,
      "learning_rate": 0.00015269825221294098,
      "loss": 0.3497,
      "step": 969
    },
    {
      "epoch": 0.9702728892501016,
      "grad_norm": 0.2983670234680176,
      "learning_rate": 0.00015260913536913154,
      "loss": 0.4093,
      "step": 970
    },
    {
      "epoch": 0.9712731705792254,
      "grad_norm": 0.3358595371246338,
      "learning_rate": 0.00015251996071739664,
      "loss": 0.3929,
      "step": 971
    },
    {
      "epoch": 0.9722734519083492,
      "grad_norm": 0.30361759662628174,
      "learning_rate": 0.00015243072835572318,
      "loss": 0.3332,
      "step": 972
    },
    {
      "epoch": 0.973273733237473,
      "grad_norm": 0.43213939666748047,
      "learning_rate": 0.0001523414383821613,
      "loss": 0.3632,
      "step": 973
    },
    {
      "epoch": 0.9742740145665969,
      "grad_norm": 0.44055286049842834,
      "learning_rate": 0.00015225209089482462,
      "loss": 0.3449,
      "step": 974
    },
    {
      "epoch": 0.9752742958957207,
      "grad_norm": 0.35058167576789856,
      "learning_rate": 0.0001521626859918898,
      "loss": 0.3633,
      "step": 975
    },
    {
      "epoch": 0.9762745772248445,
      "grad_norm": 0.4666939973831177,
      "learning_rate": 0.00015207322377159668,
      "loss": 0.3907,
      "step": 976
    },
    {
      "epoch": 0.9772748585539683,
      "grad_norm": 0.29558292031288147,
      "learning_rate": 0.00015198370433224805,
      "loss": 0.3322,
      "step": 977
    },
    {
      "epoch": 0.9782751398830921,
      "grad_norm": 0.33956781029701233,
      "learning_rate": 0.00015189412777220958,
      "loss": 0.3981,
      "step": 978
    },
    {
      "epoch": 0.9792754212122159,
      "grad_norm": 0.29612642526626587,
      "learning_rate": 0.00015180449418990976,
      "loss": 0.3773,
      "step": 979
    },
    {
      "epoch": 0.9802757025413398,
      "grad_norm": 0.3045704662799835,
      "learning_rate": 0.00015171480368383964,
      "loss": 0.3596,
      "step": 980
    },
    {
      "epoch": 0.9812759838704636,
      "grad_norm": 0.33838218450546265,
      "learning_rate": 0.00015162505635255287,
      "loss": 0.3692,
      "step": 981
    },
    {
      "epoch": 0.9822762651995874,
      "grad_norm": 0.42906007170677185,
      "learning_rate": 0.00015153525229466555,
      "loss": 0.3488,
      "step": 982
    },
    {
      "epoch": 0.9832765465287112,
      "grad_norm": 0.3778499662876129,
      "learning_rate": 0.00015144539160885613,
      "loss": 0.3257,
      "step": 983
    },
    {
      "epoch": 0.984276827857835,
      "grad_norm": 0.37084269523620605,
      "learning_rate": 0.00015135547439386516,
      "loss": 0.4242,
      "step": 984
    },
    {
      "epoch": 0.9852771091869589,
      "grad_norm": 0.3136231005191803,
      "learning_rate": 0.0001512655007484955,
      "loss": 0.3405,
      "step": 985
    },
    {
      "epoch": 0.9862773905160827,
      "grad_norm": 0.31304821372032166,
      "learning_rate": 0.00015117547077161185,
      "loss": 0.3984,
      "step": 986
    },
    {
      "epoch": 0.9872776718452064,
      "grad_norm": 0.49967846274375916,
      "learning_rate": 0.0001510853845621409,
      "loss": 0.3152,
      "step": 987
    },
    {
      "epoch": 0.9882779531743303,
      "grad_norm": 0.30760419368743896,
      "learning_rate": 0.00015099524221907107,
      "loss": 0.3703,
      "step": 988
    },
    {
      "epoch": 0.9892782345034541,
      "grad_norm": 0.3336239457130432,
      "learning_rate": 0.0001509050438414525,
      "loss": 0.3703,
      "step": 989
    },
    {
      "epoch": 0.9902785158325779,
      "grad_norm": 0.3483596742153168,
      "learning_rate": 0.00015081478952839693,
      "loss": 0.3797,
      "step": 990
    },
    {
      "epoch": 0.9912787971617018,
      "grad_norm": 0.33493107557296753,
      "learning_rate": 0.00015072447937907753,
      "loss": 0.3856,
      "step": 991
    },
    {
      "epoch": 0.9922790784908255,
      "grad_norm": 0.3484514653682709,
      "learning_rate": 0.00015063411349272877,
      "loss": 0.3613,
      "step": 992
    },
    {
      "epoch": 0.9932793598199494,
      "grad_norm": 0.2714597284793854,
      "learning_rate": 0.00015054369196864644,
      "loss": 0.3007,
      "step": 993
    },
    {
      "epoch": 0.9942796411490732,
      "grad_norm": 0.3214651644229889,
      "learning_rate": 0.00015045321490618748,
      "loss": 0.3741,
      "step": 994
    },
    {
      "epoch": 0.995279922478197,
      "grad_norm": 0.31645867228507996,
      "learning_rate": 0.00015036268240476978,
      "loss": 0.3589,
      "step": 995
    },
    {
      "epoch": 0.9962802038073209,
      "grad_norm": 0.3361944556236267,
      "learning_rate": 0.00015027209456387218,
      "loss": 0.3291,
      "step": 996
    },
    {
      "epoch": 0.9972804851364446,
      "grad_norm": 0.39736321568489075,
      "learning_rate": 0.00015018145148303438,
      "loss": 0.3377,
      "step": 997
    },
    {
      "epoch": 0.9982807664655684,
      "grad_norm": 0.39633849263191223,
      "learning_rate": 0.00015009075326185667,
      "loss": 0.4481,
      "step": 998
    },
    {
      "epoch": 0.9992810477946923,
      "grad_norm": 0.30362385511398315,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.3214,
      "step": 999
    },
    {
      "epoch": 1.0002813291238162,
      "grad_norm": 0.2802581191062927,
      "learning_rate": 0.00014990919179718584,
      "loss": 0.3184,
      "step": 1000
    }
  ],
  "logging_steps": 1,
  "max_steps": 2997,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0715904762200064e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}