{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0005626582476324,
  "eval_steps": 500,
  "global_step": 2000,
  "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
    },
    {
      "epoch": 1.0012816104529398,
      "grad_norm": 0.26813048124313354,
      "learning_rate": 0.00014981832875319597,
      "loss": 0.3136,
      "step": 1001
    },
    {
      "epoch": 1.0022818917820637,
      "grad_norm": 0.28056421875953674,
      "learning_rate": 0.00014972741096787242,
      "loss": 0.3002,
      "step": 1002
    },
    {
      "epoch": 1.0032821731111876,
      "grad_norm": 0.2766055166721344,
      "learning_rate": 0.0001496364385411174,
      "loss": 0.287,
      "step": 1003
    },
    {
      "epoch": 1.0042824544403113,
      "grad_norm": 0.3634757697582245,
      "learning_rate": 0.0001495454115728932,
      "loss": 0.3053,
      "step": 1004
    },
    {
      "epoch": 1.0052827357694352,
      "grad_norm": 0.28869709372520447,
      "learning_rate": 0.0001494543301632219,
      "loss": 0.3338,
      "step": 1005
    },
    {
      "epoch": 1.006283017098559,
      "grad_norm": 0.2862350344657898,
      "learning_rate": 0.00014936319441218555,
      "loss": 0.3127,
      "step": 1006
    },
    {
      "epoch": 1.0072832984276827,
      "grad_norm": 0.32212525606155396,
      "learning_rate": 0.0001492720044199259,
      "loss": 0.3195,
      "step": 1007
    },
    {
      "epoch": 1.0082835797568066,
      "grad_norm": 0.5240365266799927,
      "learning_rate": 0.0001491807602866442,
      "loss": 0.2763,
      "step": 1008
    },
    {
      "epoch": 1.0092838610859305,
      "grad_norm": 0.3238691985607147,
      "learning_rate": 0.00014908946211260123,
      "loss": 0.3209,
      "step": 1009
    },
    {
      "epoch": 1.0102841424150542,
      "grad_norm": 0.3728702962398529,
      "learning_rate": 0.00014899810999811726,
      "loss": 0.3139,
      "step": 1010
    },
    {
      "epoch": 1.011284423744178,
      "grad_norm": 0.347292959690094,
      "learning_rate": 0.0001489067040435717,
      "loss": 0.3048,
      "step": 1011
    },
    {
      "epoch": 1.012284705073302,
      "grad_norm": 0.3030545115470886,
      "learning_rate": 0.00014881524434940313,
      "loss": 0.3578,
      "step": 1012
    },
    {
      "epoch": 1.0132849864024256,
      "grad_norm": 0.302583247423172,
      "learning_rate": 0.0001487237310161093,
      "loss": 0.3421,
      "step": 1013
    },
    {
      "epoch": 1.0142852677315495,
      "grad_norm": 0.3362748622894287,
      "learning_rate": 0.0001486321641442467,
      "loss": 0.2959,
      "step": 1014
    },
    {
      "epoch": 1.0152855490606734,
      "grad_norm": 0.6158180236816406,
      "learning_rate": 0.00014854054383443081,
      "loss": 0.3487,
      "step": 1015
    },
    {
      "epoch": 1.016285830389797,
      "grad_norm": 0.4109620451927185,
      "learning_rate": 0.00014844887018733582,
      "loss": 0.3414,
      "step": 1016
    },
    {
      "epoch": 1.017286111718921,
      "grad_norm": 0.4342152774333954,
      "learning_rate": 0.00014835714330369446,
      "loss": 0.3386,
      "step": 1017
    },
    {
      "epoch": 1.0182863930480448,
      "grad_norm": 0.3512173891067505,
      "learning_rate": 0.00014826536328429795,
      "loss": 0.3545,
      "step": 1018
    },
    {
      "epoch": 1.0192866743771685,
      "grad_norm": 0.517636239528656,
      "learning_rate": 0.000148173530229996,
      "loss": 0.3216,
      "step": 1019
    },
    {
      "epoch": 1.0202869557062924,
      "grad_norm": 0.39727723598480225,
      "learning_rate": 0.00014808164424169647,
      "loss": 0.3224,
      "step": 1020
    },
    {
      "epoch": 1.0212872370354162,
      "grad_norm": 0.28857290744781494,
      "learning_rate": 0.0001479897054203655,
      "loss": 0.2904,
      "step": 1021
    },
    {
      "epoch": 1.0222875183645401,
      "grad_norm": 0.49946990609169006,
      "learning_rate": 0.00014789771386702717,
      "loss": 0.305,
      "step": 1022
    },
    {
      "epoch": 1.0232877996936638,
      "grad_norm": 0.25059613585472107,
      "learning_rate": 0.0001478056696827636,
      "loss": 0.272,
      "step": 1023
    },
    {
      "epoch": 1.0242880810227877,
      "grad_norm": 0.3168584406375885,
      "learning_rate": 0.0001477135729687147,
      "loss": 0.3311,
      "step": 1024
    },
    {
      "epoch": 1.0252883623519116,
      "grad_norm": 0.3028864562511444,
      "learning_rate": 0.0001476214238260781,
      "loss": 0.3187,
      "step": 1025
    },
    {
      "epoch": 1.0262886436810352,
      "grad_norm": 0.2916858196258545,
      "learning_rate": 0.000147529222356109,
      "loss": 0.2961,
      "step": 1026
    },
    {
      "epoch": 1.0272889250101591,
      "grad_norm": 0.3912205994129181,
      "learning_rate": 0.0001474369686601202,
      "loss": 0.3471,
      "step": 1027
    },
    {
      "epoch": 1.028289206339283,
      "grad_norm": 0.3413138687610626,
      "learning_rate": 0.0001473446628394818,
      "loss": 0.3238,
      "step": 1028
    },
    {
      "epoch": 1.0292894876684067,
      "grad_norm": 0.392063707113266,
      "learning_rate": 0.00014725230499562119,
      "loss": 0.3523,
      "step": 1029
    },
    {
      "epoch": 1.0302897689975306,
      "grad_norm": 0.3644559383392334,
      "learning_rate": 0.00014715989523002296,
      "loss": 0.3166,
      "step": 1030
    },
    {
      "epoch": 1.0312900503266544,
      "grad_norm": 0.35915613174438477,
      "learning_rate": 0.00014706743364422878,
      "loss": 0.3033,
      "step": 1031
    },
    {
      "epoch": 1.032290331655778,
      "grad_norm": 0.5520173907279968,
      "learning_rate": 0.00014697492033983707,
      "loss": 0.2976,
      "step": 1032
    },
    {
      "epoch": 1.033290612984902,
      "grad_norm": 0.37517377734184265,
      "learning_rate": 0.00014688235541850337,
      "loss": 0.3066,
      "step": 1033
    },
    {
      "epoch": 1.0342908943140259,
      "grad_norm": 0.37406793236732483,
      "learning_rate": 0.0001467897389819397,
      "loss": 0.3119,
      "step": 1034
    },
    {
      "epoch": 1.0352911756431495,
      "grad_norm": 0.3409208059310913,
      "learning_rate": 0.00014669707113191483,
      "loss": 0.3322,
      "step": 1035
    },
    {
      "epoch": 1.0362914569722734,
      "grad_norm": 0.30056488513946533,
      "learning_rate": 0.0001466043519702539,
      "loss": 0.3043,
      "step": 1036
    },
    {
      "epoch": 1.0372917383013973,
      "grad_norm": 0.28252458572387695,
      "learning_rate": 0.00014651158159883855,
      "loss": 0.2854,
      "step": 1037
    },
    {
      "epoch": 1.038292019630521,
      "grad_norm": 0.3065401315689087,
      "learning_rate": 0.0001464187601196066,
      "loss": 0.3258,
      "step": 1038
    },
    {
      "epoch": 1.0392923009596449,
      "grad_norm": 0.2960951328277588,
      "learning_rate": 0.00014632588763455212,
      "loss": 0.3043,
      "step": 1039
    },
    {
      "epoch": 1.0402925822887688,
      "grad_norm": 0.37415993213653564,
      "learning_rate": 0.00014623296424572517,
      "loss": 0.3025,
      "step": 1040
    },
    {
      "epoch": 1.0412928636178926,
      "grad_norm": 0.4868111312389374,
      "learning_rate": 0.00014613999005523174,
      "loss": 0.3456,
      "step": 1041
    },
    {
      "epoch": 1.0422931449470163,
      "grad_norm": 0.34748536348342896,
      "learning_rate": 0.00014604696516523361,
      "loss": 0.3465,
      "step": 1042
    },
    {
      "epoch": 1.0432934262761402,
      "grad_norm": 0.2968042492866516,
      "learning_rate": 0.00014595388967794835,
      "loss": 0.3132,
      "step": 1043
    },
    {
      "epoch": 1.044293707605264,
      "grad_norm": 0.3004485070705414,
      "learning_rate": 0.00014586076369564908,
      "loss": 0.3484,
      "step": 1044
    },
    {
      "epoch": 1.0452939889343877,
      "grad_norm": 0.32666173577308655,
      "learning_rate": 0.00014576758732066442,
      "loss": 0.2953,
      "step": 1045
    },
    {
      "epoch": 1.0462942702635116,
      "grad_norm": 0.324770987033844,
      "learning_rate": 0.00014567436065537835,
      "loss": 0.3274,
      "step": 1046
    },
    {
      "epoch": 1.0472945515926355,
      "grad_norm": 0.3392276465892792,
      "learning_rate": 0.00014558108380223012,
      "loss": 0.2994,
      "step": 1047
    },
    {
      "epoch": 1.0482948329217592,
      "grad_norm": 0.29332417249679565,
      "learning_rate": 0.00014548775686371412,
      "loss": 0.2725,
      "step": 1048
    },
    {
      "epoch": 1.049295114250883,
      "grad_norm": 0.30619457364082336,
      "learning_rate": 0.00014539437994237977,
      "loss": 0.3289,
      "step": 1049
    },
    {
      "epoch": 1.050295395580007,
      "grad_norm": 0.40373507142066956,
      "learning_rate": 0.00014530095314083143,
      "loss": 0.2716,
      "step": 1050
    },
    {
      "epoch": 1.0512956769091306,
      "grad_norm": 0.2841174900531769,
      "learning_rate": 0.00014520747656172824,
      "loss": 0.301,
      "step": 1051
    },
    {
      "epoch": 1.0522959582382545,
      "grad_norm": 0.33033978939056396,
      "learning_rate": 0.00014511395030778406,
      "loss": 0.2903,
      "step": 1052
    },
    {
      "epoch": 1.0532962395673784,
      "grad_norm": 0.29360002279281616,
      "learning_rate": 0.00014502037448176734,
      "loss": 0.2925,
      "step": 1053
    },
    {
      "epoch": 1.054296520896502,
      "grad_norm": 0.29821011424064636,
      "learning_rate": 0.000144926749186501,
      "loss": 0.3395,
      "step": 1054
    },
    {
      "epoch": 1.055296802225626,
      "grad_norm": 0.5154512524604797,
      "learning_rate": 0.00014483307452486227,
      "loss": 0.3065,
      "step": 1055
    },
    {
      "epoch": 1.0562970835547498,
      "grad_norm": 0.29638105630874634,
      "learning_rate": 0.0001447393505997827,
      "loss": 0.3418,
      "step": 1056
    },
    {
      "epoch": 1.0572973648838735,
      "grad_norm": 0.2956211268901825,
      "learning_rate": 0.00014464557751424793,
      "loss": 0.2947,
      "step": 1057
    },
    {
      "epoch": 1.0582976462129974,
      "grad_norm": 0.3215060830116272,
      "learning_rate": 0.00014455175537129758,
      "loss": 0.2892,
      "step": 1058
    },
    {
      "epoch": 1.0592979275421213,
      "grad_norm": 0.30396702885627747,
      "learning_rate": 0.00014445788427402528,
      "loss": 0.2697,
      "step": 1059
    },
    {
      "epoch": 1.0602982088712452,
      "grad_norm": 0.3325878977775574,
      "learning_rate": 0.00014436396432557835,
      "loss": 0.3076,
      "step": 1060
    },
    {
      "epoch": 1.0612984902003688,
      "grad_norm": 0.9854658842086792,
      "learning_rate": 0.00014426999562915782,
      "loss": 0.3053,
      "step": 1061
    },
    {
      "epoch": 1.0622987715294927,
      "grad_norm": 0.8612474203109741,
      "learning_rate": 0.00014417597828801832,
      "loss": 0.3357,
      "step": 1062
    },
    {
      "epoch": 1.0632990528586166,
      "grad_norm": 0.43832653760910034,
      "learning_rate": 0.0001440819124054679,
      "loss": 0.3966,
      "step": 1063
    },
    {
      "epoch": 1.0642993341877403,
      "grad_norm": 0.5110614895820618,
      "learning_rate": 0.00014398779808486793,
      "loss": 0.2765,
      "step": 1064
    },
    {
      "epoch": 1.0652996155168641,
      "grad_norm": 0.2720482051372528,
      "learning_rate": 0.00014389363542963306,
      "loss": 0.3012,
      "step": 1065
    },
    {
      "epoch": 1.066299896845988,
      "grad_norm": 0.36876875162124634,
      "learning_rate": 0.000143799424543231,
      "loss": 0.2987,
      "step": 1066
    },
    {
      "epoch": 1.0673001781751117,
      "grad_norm": 0.3124391734600067,
      "learning_rate": 0.0001437051655291825,
      "loss": 0.3072,
      "step": 1067
    },
    {
      "epoch": 1.0683004595042356,
      "grad_norm": 0.35117027163505554,
      "learning_rate": 0.0001436108584910611,
      "loss": 0.2855,
      "step": 1068
    },
    {
      "epoch": 1.0693007408333595,
      "grad_norm": 0.338226318359375,
      "learning_rate": 0.0001435165035324933,
      "loss": 0.304,
      "step": 1069
    },
    {
      "epoch": 1.0703010221624831,
      "grad_norm": 0.40764155983924866,
      "learning_rate": 0.000143422100757158,
      "loss": 0.3444,
      "step": 1070
    },
    {
      "epoch": 1.071301303491607,
      "grad_norm": 0.2941386103630066,
      "learning_rate": 0.00014332765026878687,
      "loss": 0.3092,
      "step": 1071
    },
    {
      "epoch": 1.072301584820731,
      "grad_norm": 0.30436965823173523,
      "learning_rate": 0.0001432331521711639,
      "loss": 0.3377,
      "step": 1072
    },
    {
      "epoch": 1.0733018661498546,
      "grad_norm": 0.32086703181266785,
      "learning_rate": 0.00014313860656812536,
      "loss": 0.3105,
      "step": 1073
    },
    {
      "epoch": 1.0743021474789785,
      "grad_norm": 0.3086099922657013,
      "learning_rate": 0.00014304401356355983,
      "loss": 0.3352,
      "step": 1074
    },
    {
      "epoch": 1.0753024288081023,
      "grad_norm": 0.2971747815608978,
      "learning_rate": 0.00014294937326140788,
      "loss": 0.3199,
      "step": 1075
    },
    {
      "epoch": 1.076302710137226,
      "grad_norm": 0.30569085478782654,
      "learning_rate": 0.00014285468576566207,
      "loss": 0.2952,
      "step": 1076
    },
    {
      "epoch": 1.07730299146635,
      "grad_norm": 0.34758979082107544,
      "learning_rate": 0.00014275995118036693,
      "loss": 0.2769,
      "step": 1077
    },
    {
      "epoch": 1.0783032727954738,
      "grad_norm": 0.31296706199645996,
      "learning_rate": 0.00014266516960961852,
      "loss": 0.2892,
      "step": 1078
    },
    {
      "epoch": 1.0793035541245974,
      "grad_norm": 0.32583194971084595,
      "learning_rate": 0.00014257034115756472,
      "loss": 0.3133,
      "step": 1079
    },
    {
      "epoch": 1.0803038354537213,
      "grad_norm": 0.37974250316619873,
      "learning_rate": 0.0001424754659284048,
      "loss": 0.3484,
      "step": 1080
    },
    {
      "epoch": 1.0813041167828452,
      "grad_norm": 0.29222333431243896,
      "learning_rate": 0.0001423805440263895,
      "loss": 0.3213,
      "step": 1081
    },
    {
      "epoch": 1.0823043981119689,
      "grad_norm": 0.27486810088157654,
      "learning_rate": 0.0001422855755558208,
      "loss": 0.3011,
      "step": 1082
    },
    {
      "epoch": 1.0833046794410928,
      "grad_norm": 0.3050113618373871,
      "learning_rate": 0.00014219056062105193,
      "loss": 0.3113,
      "step": 1083
    },
    {
      "epoch": 1.0843049607702167,
      "grad_norm": 0.314631849527359,
      "learning_rate": 0.0001420954993264871,
      "loss": 0.2558,
      "step": 1084
    },
    {
      "epoch": 1.0853052420993405,
      "grad_norm": 0.3079819083213806,
      "learning_rate": 0.00014200039177658145,
      "loss": 0.2893,
      "step": 1085
    },
    {
      "epoch": 1.0863055234284642,
      "grad_norm": 0.29098713397979736,
      "learning_rate": 0.000141905238075841,
      "loss": 0.3229,
      "step": 1086
    },
    {
      "epoch": 1.087305804757588,
      "grad_norm": 0.27850237488746643,
      "learning_rate": 0.00014181003832882248,
      "loss": 0.2713,
      "step": 1087
    },
    {
      "epoch": 1.088306086086712,
      "grad_norm": 0.29136067628860474,
      "learning_rate": 0.00014171479264013311,
      "loss": 0.3161,
      "step": 1088
    },
    {
      "epoch": 1.0893063674158356,
      "grad_norm": 0.2870387136936188,
      "learning_rate": 0.00014161950111443077,
      "loss": 0.2908,
      "step": 1089
    },
    {
      "epoch": 1.0903066487449595,
      "grad_norm": 0.3695163428783417,
      "learning_rate": 0.00014152416385642357,
      "loss": 0.3391,
      "step": 1090
    },
    {
      "epoch": 1.0913069300740834,
      "grad_norm": 0.3627532124519348,
      "learning_rate": 0.00014142878097086995,
      "loss": 0.3622,
      "step": 1091
    },
    {
      "epoch": 1.092307211403207,
      "grad_norm": 0.3881353735923767,
      "learning_rate": 0.0001413333525625784,
      "loss": 0.3012,
      "step": 1092
    },
    {
      "epoch": 1.093307492732331,
      "grad_norm": 0.4116741716861725,
      "learning_rate": 0.00014123787873640754,
      "loss": 0.2873,
      "step": 1093
    },
    {
      "epoch": 1.0943077740614549,
      "grad_norm": 0.5444630980491638,
      "learning_rate": 0.00014114235959726575,
      "loss": 0.3412,
      "step": 1094
    },
    {
      "epoch": 1.0953080553905785,
      "grad_norm": 0.3311006426811218,
      "learning_rate": 0.0001410467952501114,
      "loss": 0.3135,
      "step": 1095
    },
    {
      "epoch": 1.0963083367197024,
      "grad_norm": 0.2866511940956116,
      "learning_rate": 0.00014095118579995235,
      "loss": 0.3173,
      "step": 1096
    },
    {
      "epoch": 1.0973086180488263,
      "grad_norm": 0.2701050043106079,
      "learning_rate": 0.0001408555313518461,
      "loss": 0.3004,
      "step": 1097
    },
    {
      "epoch": 1.09830889937795,
      "grad_norm": 0.2890082597732544,
      "learning_rate": 0.00014075983201089964,
      "loss": 0.2996,
      "step": 1098
    },
    {
      "epoch": 1.0993091807070738,
      "grad_norm": 0.32565227150917053,
      "learning_rate": 0.0001406640878822692,
      "loss": 0.2878,
      "step": 1099
    },
    {
      "epoch": 1.1003094620361977,
      "grad_norm": 0.26777511835098267,
      "learning_rate": 0.00014056829907116024,
      "loss": 0.2906,
      "step": 1100
    },
    {
      "epoch": 1.1013097433653214,
      "grad_norm": 0.25819867849349976,
      "learning_rate": 0.00014047246568282736,
      "loss": 0.2874,
      "step": 1101
    },
    {
      "epoch": 1.1023100246944453,
      "grad_norm": 0.3067924380302429,
      "learning_rate": 0.00014037658782257414,
      "loss": 0.3525,
      "step": 1102
    },
    {
      "epoch": 1.1033103060235692,
      "grad_norm": 0.4779090881347656,
      "learning_rate": 0.00014028066559575302,
      "loss": 0.3407,
      "step": 1103
    },
    {
      "epoch": 1.104310587352693,
      "grad_norm": 0.2718350887298584,
      "learning_rate": 0.00014018469910776513,
      "loss": 0.2783,
      "step": 1104
    },
    {
      "epoch": 1.1053108686818167,
      "grad_norm": 0.6914875507354736,
      "learning_rate": 0.0001400886884640603,
      "loss": 0.2949,
      "step": 1105
    },
    {
      "epoch": 1.1063111500109406,
      "grad_norm": 0.4071102440357208,
      "learning_rate": 0.00013999263377013693,
      "loss": 0.2838,
      "step": 1106
    },
    {
      "epoch": 1.1073114313400645,
      "grad_norm": 0.39632922410964966,
      "learning_rate": 0.00013989653513154165,
      "loss": 0.369,
      "step": 1107
    },
    {
      "epoch": 1.1083117126691882,
      "grad_norm": 0.4760614335536957,
      "learning_rate": 0.00013980039265386955,
      "loss": 0.2933,
      "step": 1108
    },
    {
      "epoch": 1.109311993998312,
      "grad_norm": 0.32944560050964355,
      "learning_rate": 0.00013970420644276383,
      "loss": 0.2744,
      "step": 1109
    },
    {
      "epoch": 1.110312275327436,
      "grad_norm": 0.5836090445518494,
      "learning_rate": 0.0001396079766039157,
      "loss": 0.2837,
      "step": 1110
    },
    {
      "epoch": 1.1113125566565596,
      "grad_norm": 0.3504053056240082,
      "learning_rate": 0.00013951170324306435,
      "loss": 0.2844,
      "step": 1111
    },
    {
      "epoch": 1.1123128379856835,
      "grad_norm": 0.3881485164165497,
      "learning_rate": 0.00013941538646599687,
      "loss": 0.2864,
      "step": 1112
    },
    {
      "epoch": 1.1133131193148074,
      "grad_norm": 0.29575201869010925,
      "learning_rate": 0.0001393190263785479,
      "loss": 0.3108,
      "step": 1113
    },
    {
      "epoch": 1.114313400643931,
      "grad_norm": 0.9517633318901062,
      "learning_rate": 0.0001392226230865998,
      "loss": 0.3617,
      "step": 1114
    },
    {
      "epoch": 1.115313681973055,
      "grad_norm": 0.28317132592201233,
      "learning_rate": 0.0001391261766960823,
      "loss": 0.2761,
      "step": 1115
    },
    {
      "epoch": 1.1163139633021788,
      "grad_norm": 0.2812111973762512,
      "learning_rate": 0.00013902968731297255,
      "loss": 0.2755,
      "step": 1116
    },
    {
      "epoch": 1.1173142446313025,
      "grad_norm": 0.3398970067501068,
      "learning_rate": 0.00013893315504329498,
      "loss": 0.3309,
      "step": 1117
    },
    {
      "epoch": 1.1183145259604264,
      "grad_norm": 0.25279682874679565,
      "learning_rate": 0.00013883657999312109,
      "loss": 0.2787,
      "step": 1118
    },
    {
      "epoch": 1.1193148072895502,
      "grad_norm": 0.3351457715034485,
      "learning_rate": 0.00013873996226856933,
      "loss": 0.3733,
      "step": 1119
    },
    {
      "epoch": 1.120315088618674,
      "grad_norm": 0.3026616871356964,
      "learning_rate": 0.00013864330197580513,
      "loss": 0.3437,
      "step": 1120
    },
    {
      "epoch": 1.1213153699477978,
      "grad_norm": 0.2620201110839844,
      "learning_rate": 0.0001385465992210407,
      "loss": 0.2693,
      "step": 1121
    },
    {
      "epoch": 1.1223156512769217,
      "grad_norm": 0.28102657198905945,
      "learning_rate": 0.00013844985411053492,
      "loss": 0.312,
      "step": 1122
    },
    {
      "epoch": 1.1233159326060456,
      "grad_norm": 0.305686354637146,
      "learning_rate": 0.00013835306675059308,
      "loss": 0.3131,
      "step": 1123
    },
    {
      "epoch": 1.1243162139351692,
      "grad_norm": 0.29613766074180603,
      "learning_rate": 0.00013825623724756704,
      "loss": 0.3134,
      "step": 1124
    },
    {
      "epoch": 1.1253164952642931,
      "grad_norm": 0.4292270839214325,
      "learning_rate": 0.00013815936570785487,
      "loss": 0.284,
      "step": 1125
    },
    {
      "epoch": 1.1263167765934168,
      "grad_norm": 0.34200307726860046,
      "learning_rate": 0.00013806245223790088,
      "loss": 0.3308,
      "step": 1126
    },
    {
      "epoch": 1.1273170579225407,
      "grad_norm": 0.28017300367355347,
      "learning_rate": 0.0001379654969441955,
      "loss": 0.3116,
      "step": 1127
    },
    {
      "epoch": 1.1283173392516646,
      "grad_norm": 0.270203173160553,
      "learning_rate": 0.000137868499933275,
      "loss": 0.3021,
      "step": 1128
    },
    {
      "epoch": 1.1293176205807884,
      "grad_norm": 0.8500344157218933,
      "learning_rate": 0.00013777146131172162,
      "loss": 0.3005,
      "step": 1129
    },
    {
      "epoch": 1.130317901909912,
      "grad_norm": 0.2777160108089447,
      "learning_rate": 0.00013767438118616318,
      "loss": 0.2826,
      "step": 1130
    },
    {
      "epoch": 1.131318183239036,
      "grad_norm": 0.3133338689804077,
      "learning_rate": 0.00013757725966327322,
      "loss": 0.2774,
      "step": 1131
    },
    {
      "epoch": 1.1323184645681599,
      "grad_norm": 0.35257449746131897,
      "learning_rate": 0.00013748009684977073,
      "loss": 0.3796,
      "step": 1132
    },
    {
      "epoch": 1.1333187458972835,
      "grad_norm": 0.33504629135131836,
      "learning_rate": 0.0001373828928524201,
      "loss": 0.3019,
      "step": 1133
    },
    {
      "epoch": 1.1343190272264074,
      "grad_norm": 0.3297973573207855,
      "learning_rate": 0.00013728564777803088,
      "loss": 0.314,
      "step": 1134
    },
    {
      "epoch": 1.1353193085555313,
      "grad_norm": 0.2501577138900757,
      "learning_rate": 0.00013718836173345783,
      "loss": 0.2692,
      "step": 1135
    },
    {
      "epoch": 1.136319589884655,
      "grad_norm": 0.2842780351638794,
      "learning_rate": 0.00013709103482560078,
      "loss": 0.3072,
      "step": 1136
    },
    {
      "epoch": 1.1373198712137789,
      "grad_norm": 0.3117831349372864,
      "learning_rate": 0.00013699366716140435,
      "loss": 0.2771,
      "step": 1137
    },
    {
      "epoch": 1.1383201525429028,
      "grad_norm": 0.3196011483669281,
      "learning_rate": 0.00013689625884785798,
      "loss": 0.2803,
      "step": 1138
    },
    {
      "epoch": 1.1393204338720264,
      "grad_norm": 0.31798556447029114,
      "learning_rate": 0.00013679880999199583,
      "loss": 0.3189,
      "step": 1139
    },
    {
      "epoch": 1.1403207152011503,
      "grad_norm": 0.37045398354530334,
      "learning_rate": 0.00013670132070089653,
      "loss": 0.3457,
      "step": 1140
    },
    {
      "epoch": 1.1413209965302742,
      "grad_norm": 0.28372254967689514,
      "learning_rate": 0.00013660379108168324,
      "loss": 0.2749,
      "step": 1141
    },
    {
      "epoch": 1.142321277859398,
      "grad_norm": 0.393475741147995,
      "learning_rate": 0.00013650622124152334,
      "loss": 0.314,
      "step": 1142
    },
    {
      "epoch": 1.1433215591885217,
      "grad_norm": 0.3403913676738739,
      "learning_rate": 0.0001364086112876284,
      "loss": 0.3303,
      "step": 1143
    },
    {
      "epoch": 1.1443218405176456,
      "grad_norm": 0.32799428701400757,
      "learning_rate": 0.00013631096132725413,
      "loss": 0.3095,
      "step": 1144
    },
    {
      "epoch": 1.1453221218467693,
      "grad_norm": 0.3379594683647156,
      "learning_rate": 0.00013621327146770025,
      "loss": 0.3287,
      "step": 1145
    },
    {
      "epoch": 1.1463224031758932,
      "grad_norm": 0.4616290330886841,
      "learning_rate": 0.00013611554181631013,
      "loss": 0.3212,
      "step": 1146
    },
    {
      "epoch": 1.147322684505017,
      "grad_norm": 0.3926517963409424,
      "learning_rate": 0.00013601777248047105,
      "loss": 0.3565,
      "step": 1147
    },
    {
      "epoch": 1.148322965834141,
      "grad_norm": 0.5697782635688782,
      "learning_rate": 0.0001359199635676138,
      "loss": 0.2964,
      "step": 1148
    },
    {
      "epoch": 1.1493232471632646,
      "grad_norm": 0.33533936738967896,
      "learning_rate": 0.00013582211518521273,
      "loss": 0.2873,
      "step": 1149
    },
    {
      "epoch": 1.1503235284923885,
      "grad_norm": 0.394722044467926,
      "learning_rate": 0.00013572422744078551,
      "loss": 0.3016,
      "step": 1150
    },
    {
      "epoch": 1.1513238098215124,
      "grad_norm": 0.28595203161239624,
      "learning_rate": 0.00013562630044189304,
      "loss": 0.3196,
      "step": 1151
    },
    {
      "epoch": 1.152324091150636,
      "grad_norm": 0.31986042857170105,
      "learning_rate": 0.00013552833429613938,
      "loss": 0.3156,
      "step": 1152
    },
    {
      "epoch": 1.15332437247976,
      "grad_norm": 0.26666995882987976,
      "learning_rate": 0.0001354303291111716,
      "loss": 0.3003,
      "step": 1153
    },
    {
      "epoch": 1.1543246538088838,
      "grad_norm": 0.281917929649353,
      "learning_rate": 0.0001353322849946797,
      "loss": 0.2989,
      "step": 1154
    },
    {
      "epoch": 1.1553249351380075,
      "grad_norm": 0.3064289093017578,
      "learning_rate": 0.00013523420205439646,
      "loss": 0.3313,
      "step": 1155
    },
    {
      "epoch": 1.1563252164671314,
      "grad_norm": 0.31628215312957764,
      "learning_rate": 0.0001351360803980972,
      "loss": 0.342,
      "step": 1156
    },
    {
      "epoch": 1.1573254977962553,
      "grad_norm": 0.3470976650714874,
      "learning_rate": 0.00013503792013359997,
      "loss": 0.3333,
      "step": 1157
    },
    {
      "epoch": 1.158325779125379,
      "grad_norm": 0.3361537456512451,
      "learning_rate": 0.00013493972136876509,
      "loss": 0.2875,
      "step": 1158
    },
    {
      "epoch": 1.1593260604545028,
      "grad_norm": 0.3168264329433441,
      "learning_rate": 0.00013484148421149527,
      "loss": 0.3084,
      "step": 1159
    },
    {
      "epoch": 1.1603263417836267,
      "grad_norm": 0.3275001347064972,
      "learning_rate": 0.0001347432087697354,
      "loss": 0.2964,
      "step": 1160
    },
    {
      "epoch": 1.1613266231127504,
      "grad_norm": 0.2674354910850525,
      "learning_rate": 0.00013464489515147238,
      "loss": 0.2817,
      "step": 1161
    },
    {
      "epoch": 1.1623269044418743,
      "grad_norm": 0.2902231514453888,
      "learning_rate": 0.0001345465434647351,
      "loss": 0.3459,
      "step": 1162
    },
    {
      "epoch": 1.1633271857709981,
      "grad_norm": 0.3231171667575836,
      "learning_rate": 0.00013444815381759425,
      "loss": 0.2603,
      "step": 1163
    },
    {
      "epoch": 1.1643274671001218,
      "grad_norm": 0.26167550683021545,
      "learning_rate": 0.00013434972631816235,
      "loss": 0.253,
      "step": 1164
    },
    {
      "epoch": 1.1653277484292457,
      "grad_norm": 0.2913510501384735,
      "learning_rate": 0.0001342512610745933,
      "loss": 0.333,
      "step": 1165
    },
    {
      "epoch": 1.1663280297583696,
      "grad_norm": 0.30092528462409973,
      "learning_rate": 0.0001341527581950827,
      "loss": 0.2871,
      "step": 1166
    },
    {
      "epoch": 1.1673283110874935,
      "grad_norm": 1.1661570072174072,
      "learning_rate": 0.00013405421778786737,
      "loss": 0.3051,
      "step": 1167
    },
    {
      "epoch": 1.1683285924166171,
      "grad_norm": 0.42999863624572754,
      "learning_rate": 0.00013395563996122537,
      "loss": 0.2779,
      "step": 1168
    },
    {
      "epoch": 1.169328873745741,
      "grad_norm": 0.4190734922885895,
      "learning_rate": 0.00013385702482347593,
      "loss": 0.2972,
      "step": 1169
    },
    {
      "epoch": 1.170329155074865,
      "grad_norm": 0.4557470679283142,
      "learning_rate": 0.00013375837248297926,
      "loss": 0.2562,
      "step": 1170
    },
    {
      "epoch": 1.1713294364039886,
      "grad_norm": 0.35442307591438293,
      "learning_rate": 0.0001336596830481364,
      "loss": 0.2993,
      "step": 1171
    },
    {
      "epoch": 1.1723297177331125,
      "grad_norm": 0.6263668537139893,
      "learning_rate": 0.0001335609566273892,
      "loss": 0.3277,
      "step": 1172
    },
    {
      "epoch": 1.1733299990622363,
      "grad_norm": 0.3042425811290741,
      "learning_rate": 0.00013346219332922016,
      "loss": 0.3273,
      "step": 1173
    },
    {
      "epoch": 1.17433028039136,
      "grad_norm": 0.2900557816028595,
      "learning_rate": 0.00013336339326215228,
      "loss": 0.266,
      "step": 1174
    },
    {
      "epoch": 1.175330561720484,
      "grad_norm": 0.3598930239677429,
      "learning_rate": 0.00013326455653474897,
      "loss": 0.2961,
      "step": 1175
    },
    {
      "epoch": 1.1763308430496078,
      "grad_norm": 0.31951040029525757,
      "learning_rate": 0.00013316568325561393,
      "loss": 0.2956,
      "step": 1176
    },
    {
      "epoch": 1.1773311243787314,
      "grad_norm": 0.4260864853858948,
      "learning_rate": 0.00013306677353339098,
      "loss": 0.3342,
      "step": 1177
    },
    {
      "epoch": 1.1783314057078553,
      "grad_norm": 0.2966102063655853,
      "learning_rate": 0.000132967827476764,
      "loss": 0.2782,
      "step": 1178
    },
    {
      "epoch": 1.1793316870369792,
      "grad_norm": 0.28297337889671326,
      "learning_rate": 0.0001328688451944569,
      "loss": 0.2605,
      "step": 1179
    },
    {
      "epoch": 1.1803319683661029,
      "grad_norm": 0.2880258858203888,
      "learning_rate": 0.00013276982679523322,
      "loss": 0.2914,
      "step": 1180
    },
    {
      "epoch": 1.1813322496952268,
      "grad_norm": 0.28763651847839355,
      "learning_rate": 0.00013267077238789633,
      "loss": 0.2877,
      "step": 1181
    },
    {
      "epoch": 1.1823325310243507,
      "grad_norm": 0.2898299992084503,
      "learning_rate": 0.00013257168208128908,
      "loss": 0.3203,
      "step": 1182
    },
    {
      "epoch": 1.1833328123534743,
      "grad_norm": 0.5219337940216064,
      "learning_rate": 0.00013247255598429378,
      "loss": 0.3236,
      "step": 1183
    },
    {
      "epoch": 1.1843330936825982,
      "grad_norm": 0.3025731146335602,
      "learning_rate": 0.00013237339420583212,
      "loss": 0.3019,
      "step": 1184
    },
    {
      "epoch": 1.185333375011722,
      "grad_norm": 0.32032182812690735,
      "learning_rate": 0.00013227419685486492,
      "loss": 0.2709,
      "step": 1185
    },
    {
      "epoch": 1.186333656340846,
      "grad_norm": 0.3330485224723816,
      "learning_rate": 0.00013217496404039218,
      "loss": 0.3264,
      "step": 1186
    },
    {
      "epoch": 1.1873339376699696,
      "grad_norm": 0.3723902404308319,
      "learning_rate": 0.0001320756958714528,
      "loss": 0.3423,
      "step": 1187
    },
    {
      "epoch": 1.1883342189990935,
      "grad_norm": 0.3019760251045227,
      "learning_rate": 0.00013197639245712454,
      "loss": 0.2777,
      "step": 1188
    },
    {
      "epoch": 1.1893345003282172,
      "grad_norm": 0.3286869525909424,
      "learning_rate": 0.00013187705390652388,
      "loss": 0.2952,
      "step": 1189
    },
    {
      "epoch": 1.190334781657341,
      "grad_norm": 0.4110366702079773,
      "learning_rate": 0.00013177768032880593,
      "loss": 0.2996,
      "step": 1190
    },
    {
      "epoch": 1.191335062986465,
      "grad_norm": 0.4172464609146118,
      "learning_rate": 0.0001316782718331643,
      "loss": 0.3116,
      "step": 1191
    },
    {
      "epoch": 1.1923353443155889,
      "grad_norm": 0.41905805468559265,
      "learning_rate": 0.0001315788285288309,
      "loss": 0.3307,
      "step": 1192
    },
    {
      "epoch": 1.1933356256447125,
      "grad_norm": 0.34171828627586365,
      "learning_rate": 0.00013147935052507597,
      "loss": 0.3328,
      "step": 1193
    },
    {
      "epoch": 1.1943359069738364,
      "grad_norm": 0.4480704665184021,
      "learning_rate": 0.00013137983793120786,
      "loss": 0.3014,
      "step": 1194
    },
    {
      "epoch": 1.1953361883029603,
      "grad_norm": 0.30034685134887695,
      "learning_rate": 0.0001312802908565729,
      "loss": 0.2869,
      "step": 1195
    },
    {
      "epoch": 1.196336469632084,
      "grad_norm": 0.3072358965873718,
      "learning_rate": 0.0001311807094105553,
      "loss": 0.3045,
      "step": 1196
    },
    {
      "epoch": 1.1973367509612078,
      "grad_norm": 0.337197870016098,
      "learning_rate": 0.00013108109370257712,
      "loss": 0.2925,
      "step": 1197
    },
    {
      "epoch": 1.1983370322903317,
      "grad_norm": 0.29965418577194214,
      "learning_rate": 0.00013098144384209796,
      "loss": 0.2926,
      "step": 1198
    },
    {
      "epoch": 1.1993373136194554,
      "grad_norm": 0.3737524151802063,
      "learning_rate": 0.000130881759938615,
      "loss": 0.295,
      "step": 1199
    },
    {
      "epoch": 1.2003375949485793,
      "grad_norm": 0.3686083257198334,
      "learning_rate": 0.00013078204210166278,
      "loss": 0.278,
      "step": 1200
    },
    {
      "epoch": 1.2013378762777032,
      "grad_norm": 0.3025103807449341,
      "learning_rate": 0.00013068229044081324,
      "loss": 0.3057,
      "step": 1201
    },
    {
      "epoch": 1.2023381576068268,
      "grad_norm": 0.4150354862213135,
      "learning_rate": 0.0001305825050656754,
      "loss": 0.3018,
      "step": 1202
    },
    {
      "epoch": 1.2033384389359507,
      "grad_norm": 0.29230961203575134,
      "learning_rate": 0.00013048268608589533,
      "loss": 0.2773,
      "step": 1203
    },
    {
      "epoch": 1.2043387202650746,
      "grad_norm": 0.41539573669433594,
      "learning_rate": 0.00013038283361115603,
      "loss": 0.284,
      "step": 1204
    },
    {
      "epoch": 1.2053390015941985,
      "grad_norm": 0.312111496925354,
      "learning_rate": 0.0001302829477511773,
      "loss": 0.317,
      "step": 1205
    },
    {
      "epoch": 1.2063392829233222,
      "grad_norm": 0.29801467061042786,
      "learning_rate": 0.0001301830286157157,
      "loss": 0.2757,
      "step": 1206
    },
    {
      "epoch": 1.207339564252446,
      "grad_norm": 0.31586360931396484,
      "learning_rate": 0.0001300830763145642,
      "loss": 0.2674,
      "step": 1207
    },
    {
      "epoch": 1.2083398455815697,
      "grad_norm": 0.3015751540660858,
      "learning_rate": 0.00012998309095755235,
      "loss": 0.3385,
      "step": 1208
    },
    {
      "epoch": 1.2093401269106936,
      "grad_norm": 0.2895602583885193,
      "learning_rate": 0.00012988307265454597,
      "loss": 0.3123,
      "step": 1209
    },
    {
      "epoch": 1.2103404082398175,
      "grad_norm": 0.26979225873947144,
      "learning_rate": 0.0001297830215154471,
      "loss": 0.298,
      "step": 1210
    },
    {
      "epoch": 1.2113406895689414,
      "grad_norm": 0.2885194718837738,
      "learning_rate": 0.00012968293765019384,
      "loss": 0.2771,
      "step": 1211
    },
    {
      "epoch": 1.212340970898065,
      "grad_norm": 0.3985464870929718,
      "learning_rate": 0.00012958282116876026,
      "loss": 0.354,
      "step": 1212
    },
    {
      "epoch": 1.213341252227189,
      "grad_norm": 0.303090363740921,
      "learning_rate": 0.00012948267218115624,
      "loss": 0.3126,
      "step": 1213
    },
    {
      "epoch": 1.2143415335563128,
      "grad_norm": 0.3002830147743225,
      "learning_rate": 0.00012938249079742743,
      "loss": 0.3247,
      "step": 1214
    },
    {
      "epoch": 1.2153418148854365,
      "grad_norm": 0.2839539051055908,
      "learning_rate": 0.00012928227712765504,
      "loss": 0.2952,
      "step": 1215
    },
    {
      "epoch": 1.2163420962145604,
      "grad_norm": 0.27370113134384155,
      "learning_rate": 0.0001291820312819558,
      "loss": 0.2735,
      "step": 1216
    },
    {
      "epoch": 1.2173423775436842,
      "grad_norm": 0.30879032611846924,
      "learning_rate": 0.00012908175337048174,
      "loss": 0.3237,
      "step": 1217
    },
    {
      "epoch": 1.218342658872808,
      "grad_norm": 0.27697813510894775,
      "learning_rate": 0.00012898144350342015,
      "loss": 0.2697,
      "step": 1218
    },
    {
      "epoch": 1.2193429402019318,
      "grad_norm": 0.31954845786094666,
      "learning_rate": 0.0001288811017909934,
      "loss": 0.3063,
      "step": 1219
    },
    {
      "epoch": 1.2203432215310557,
      "grad_norm": 0.32831844687461853,
      "learning_rate": 0.00012878072834345895,
      "loss": 0.305,
      "step": 1220
    },
    {
      "epoch": 1.2213435028601793,
      "grad_norm": 0.30406394600868225,
      "learning_rate": 0.00012868032327110904,
      "loss": 0.3126,
      "step": 1221
    },
    {
      "epoch": 1.2223437841893032,
      "grad_norm": 0.5284988284111023,
      "learning_rate": 0.00012857988668427066,
      "loss": 0.3382,
      "step": 1222
    },
    {
      "epoch": 1.2233440655184271,
      "grad_norm": 0.3502587378025055,
      "learning_rate": 0.0001284794186933055,
      "loss": 0.319,
      "step": 1223
    },
    {
      "epoch": 1.2243443468475508,
      "grad_norm": 0.4602774977684021,
      "learning_rate": 0.00012837891940860972,
      "loss": 0.2832,
      "step": 1224
    },
    {
      "epoch": 1.2253446281766747,
      "grad_norm": 0.2990822494029999,
      "learning_rate": 0.00012827838894061377,
      "loss": 0.2965,
      "step": 1225
    },
    {
      "epoch": 1.2263449095057986,
      "grad_norm": 0.30837929248809814,
      "learning_rate": 0.00012817782739978255,
      "loss": 0.2939,
      "step": 1226
    },
    {
      "epoch": 1.2273451908349222,
      "grad_norm": 0.38700002431869507,
      "learning_rate": 0.00012807723489661495,
      "loss": 0.3099,
      "step": 1227
    },
    {
      "epoch": 1.228345472164046,
      "grad_norm": 0.35250920057296753,
      "learning_rate": 0.00012797661154164395,
      "loss": 0.3446,
      "step": 1228
    },
    {
      "epoch": 1.22934575349317,
      "grad_norm": 0.27783432602882385,
      "learning_rate": 0.00012787595744543647,
      "loss": 0.3214,
      "step": 1229
    },
    {
      "epoch": 1.2303460348222939,
      "grad_norm": 0.31061607599258423,
      "learning_rate": 0.00012777527271859307,
      "loss": 0.3201,
      "step": 1230
    },
    {
      "epoch": 1.2313463161514175,
      "grad_norm": 0.30650797486305237,
      "learning_rate": 0.0001276745574717481,
      "loss": 0.3041,
      "step": 1231
    },
    {
      "epoch": 1.2323465974805414,
      "grad_norm": 0.32242551445961,
      "learning_rate": 0.00012757381181556943,
      "loss": 0.2925,
      "step": 1232
    },
    {
      "epoch": 1.2333468788096653,
      "grad_norm": 0.26710522174835205,
      "learning_rate": 0.0001274730358607583,
      "loss": 0.3012,
      "step": 1233
    },
    {
      "epoch": 1.234347160138789,
      "grad_norm": 0.27486371994018555,
      "learning_rate": 0.00012737222971804924,
      "loss": 0.3111,
      "step": 1234
    },
    {
      "epoch": 1.2353474414679129,
      "grad_norm": 0.2983880937099457,
      "learning_rate": 0.00012727139349821,
      "loss": 0.3362,
      "step": 1235
    },
    {
      "epoch": 1.2363477227970368,
      "grad_norm": 0.296804279088974,
      "learning_rate": 0.0001271705273120413,
      "loss": 0.2972,
      "step": 1236
    },
    {
      "epoch": 1.2373480041261604,
      "grad_norm": 0.31436458230018616,
      "learning_rate": 0.00012706963127037685,
      "loss": 0.3052,
      "step": 1237
    },
    {
      "epoch": 1.2383482854552843,
      "grad_norm": 0.3212319314479828,
      "learning_rate": 0.00012696870548408316,
      "loss": 0.2914,
      "step": 1238
    },
    {
      "epoch": 1.2393485667844082,
      "grad_norm": 0.31861424446105957,
      "learning_rate": 0.00012686775006405946,
      "loss": 0.3181,
      "step": 1239
    },
    {
      "epoch": 1.2403488481135319,
      "grad_norm": 0.27707338333129883,
      "learning_rate": 0.00012676676512123747,
      "loss": 0.2707,
      "step": 1240
    },
    {
      "epoch": 1.2413491294426557,
      "grad_norm": 0.27573227882385254,
      "learning_rate": 0.00012666575076658134,
      "loss": 0.2845,
      "step": 1241
    },
    {
      "epoch": 1.2423494107717796,
      "grad_norm": 0.28983259201049805,
      "learning_rate": 0.00012656470711108764,
      "loss": 0.3261,
      "step": 1242
    },
    {
      "epoch": 1.2433496921009033,
      "grad_norm": 0.2832695543766022,
      "learning_rate": 0.00012646363426578505,
      "loss": 0.3142,
      "step": 1243
    },
    {
      "epoch": 1.2443499734300272,
      "grad_norm": 0.4030449390411377,
      "learning_rate": 0.0001263625323417343,
      "loss": 0.273,
      "step": 1244
    },
    {
      "epoch": 1.245350254759151,
      "grad_norm": 0.30458173155784607,
      "learning_rate": 0.0001262614014500282,
      "loss": 0.3019,
      "step": 1245
    },
    {
      "epoch": 1.2463505360882747,
      "grad_norm": 0.3097798526287079,
      "learning_rate": 0.00012616024170179126,
      "loss": 0.3061,
      "step": 1246
    },
    {
      "epoch": 1.2473508174173986,
      "grad_norm": 0.32121607661247253,
      "learning_rate": 0.00012605905320817976,
      "loss": 0.3231,
      "step": 1247
    },
    {
      "epoch": 1.2483510987465225,
      "grad_norm": 0.33073854446411133,
      "learning_rate": 0.00012595783608038155,
      "loss": 0.2861,
      "step": 1248
    },
    {
      "epoch": 1.2493513800756464,
      "grad_norm": 0.36477917432785034,
      "learning_rate": 0.00012585659042961596,
      "loss": 0.3089,
      "step": 1249
    },
    {
      "epoch": 1.25035166140477,
      "grad_norm": 0.3822841942310333,
      "learning_rate": 0.00012575531636713368,
      "loss": 0.2841,
      "step": 1250
    },
    {
      "epoch": 1.251351942733894,
      "grad_norm": 0.40426018834114075,
      "learning_rate": 0.00012565401400421651,
      "loss": 0.3088,
      "step": 1251
    },
    {
      "epoch": 1.2523522240630176,
      "grad_norm": 0.28871846199035645,
      "learning_rate": 0.0001255526834521775,
      "loss": 0.3217,
      "step": 1252
    },
    {
      "epoch": 1.2533525053921415,
      "grad_norm": 0.2652817368507385,
      "learning_rate": 0.00012545132482236055,
      "loss": 0.2494,
      "step": 1253
    },
    {
      "epoch": 1.2543527867212654,
      "grad_norm": 0.28944849967956543,
      "learning_rate": 0.0001253499382261405,
      "loss": 0.2821,
      "step": 1254
    },
    {
      "epoch": 1.2553530680503893,
      "grad_norm": 0.30837482213974,
      "learning_rate": 0.00012524852377492285,
      "loss": 0.3222,
      "step": 1255
    },
    {
      "epoch": 1.256353349379513,
      "grad_norm": 0.3551330268383026,
      "learning_rate": 0.00012514708158014378,
      "loss": 0.3417,
      "step": 1256
    },
    {
      "epoch": 1.2573536307086368,
      "grad_norm": 0.2909022271633148,
      "learning_rate": 0.00012504561175326985,
      "loss": 0.2983,
      "step": 1257
    },
    {
      "epoch": 1.2583539120377607,
      "grad_norm": 0.2956262528896332,
      "learning_rate": 0.00012494411440579814,
      "loss": 0.3263,
      "step": 1258
    },
    {
      "epoch": 1.2593541933668844,
      "grad_norm": 0.49296051263809204,
      "learning_rate": 0.0001248425896492558,
      "loss": 0.342,
      "step": 1259
    },
    {
      "epoch": 1.2603544746960083,
      "grad_norm": 0.30851542949676514,
      "learning_rate": 0.00012474103759520027,
      "loss": 0.3204,
      "step": 1260
    },
    {
      "epoch": 1.2613547560251321,
      "grad_norm": 0.3080751299858093,
      "learning_rate": 0.00012463945835521878,
      "loss": 0.2874,
      "step": 1261
    },
    {
      "epoch": 1.2623550373542558,
      "grad_norm": 0.2856109142303467,
      "learning_rate": 0.0001245378520409286,
      "loss": 0.2827,
      "step": 1262
    },
    {
      "epoch": 1.2633553186833797,
      "grad_norm": 0.3182509243488312,
      "learning_rate": 0.0001244362187639767,
      "loss": 0.2898,
      "step": 1263
    },
    {
      "epoch": 1.2643556000125036,
      "grad_norm": 0.30298247933387756,
      "learning_rate": 0.00012433455863603967,
      "loss": 0.2991,
      "step": 1264
    },
    {
      "epoch": 1.2653558813416272,
      "grad_norm": 0.3058716654777527,
      "learning_rate": 0.00012423287176882358,
      "loss": 0.3583,
      "step": 1265
    },
    {
      "epoch": 1.2663561626707511,
      "grad_norm": 0.2898925244808197,
      "learning_rate": 0.00012413115827406392,
      "loss": 0.2829,
      "step": 1266
    },
    {
      "epoch": 1.267356443999875,
      "grad_norm": 0.42578408122062683,
      "learning_rate": 0.00012402941826352546,
      "loss": 0.2824,
      "step": 1267
    },
    {
      "epoch": 1.268356725328999,
      "grad_norm": 0.2729225754737854,
      "learning_rate": 0.00012392765184900202,
      "loss": 0.2789,
      "step": 1268
    },
    {
      "epoch": 1.2693570066581226,
      "grad_norm": 0.26908326148986816,
      "learning_rate": 0.0001238258591423165,
      "loss": 0.2692,
      "step": 1269
    },
    {
      "epoch": 1.2703572879872465,
      "grad_norm": 0.33527854084968567,
      "learning_rate": 0.00012372404025532072,
      "loss": 0.3218,
      "step": 1270
    },
    {
      "epoch": 1.2713575693163701,
      "grad_norm": 0.3184176981449127,
      "learning_rate": 0.00012362219529989514,
      "loss": 0.3005,
      "step": 1271
    },
    {
      "epoch": 1.272357850645494,
      "grad_norm": 0.36007922887802124,
      "learning_rate": 0.00012352032438794902,
      "loss": 0.2908,
      "step": 1272
    },
    {
      "epoch": 1.273358131974618,
      "grad_norm": 0.2578636705875397,
      "learning_rate": 0.00012341842763142005,
      "loss": 0.2707,
      "step": 1273
    },
    {
      "epoch": 1.2743584133037418,
      "grad_norm": 0.3254690170288086,
      "learning_rate": 0.00012331650514227425,
      "loss": 0.3311,
      "step": 1274
    },
    {
      "epoch": 1.2753586946328654,
      "grad_norm": 0.314831018447876,
      "learning_rate": 0.00012321455703250616,
      "loss": 0.3295,
      "step": 1275
    },
    {
      "epoch": 1.2763589759619893,
      "grad_norm": 0.29057860374450684,
      "learning_rate": 0.00012311258341413822,
      "loss": 0.2613,
      "step": 1276
    },
    {
      "epoch": 1.277359257291113,
      "grad_norm": 0.3178429901599884,
      "learning_rate": 0.00012301058439922102,
      "loss": 0.289,
      "step": 1277
    },
    {
      "epoch": 1.2783595386202369,
      "grad_norm": 0.29894882440567017,
      "learning_rate": 0.000122908560099833,
      "loss": 0.3245,
      "step": 1278
    },
    {
      "epoch": 1.2793598199493608,
      "grad_norm": 0.41435906291007996,
      "learning_rate": 0.00012280651062808047,
      "loss": 0.3319,
      "step": 1279
    },
    {
      "epoch": 1.2803601012784847,
      "grad_norm": 0.3234275281429291,
      "learning_rate": 0.00012270443609609729,
      "loss": 0.2862,
      "step": 1280
    },
    {
      "epoch": 1.2813603826076083,
      "grad_norm": 0.28792110085487366,
      "learning_rate": 0.0001226023366160449,
      "loss": 0.3081,
      "step": 1281
    },
    {
      "epoch": 1.2823606639367322,
      "grad_norm": 0.30790042877197266,
      "learning_rate": 0.00012250021230011225,
      "loss": 0.3253,
      "step": 1282
    },
    {
      "epoch": 1.283360945265856,
      "grad_norm": 0.31892886757850647,
      "learning_rate": 0.00012239806326051539,
      "loss": 0.2947,
      "step": 1283
    },
    {
      "epoch": 1.2843612265949798,
      "grad_norm": 0.31567952036857605,
      "learning_rate": 0.00012229588960949771,
      "loss": 0.3419,
      "step": 1284
    },
    {
      "epoch": 1.2853615079241036,
      "grad_norm": 0.3066472113132477,
      "learning_rate": 0.00012219369145932959,
      "loss": 0.3438,
      "step": 1285
    },
    {
      "epoch": 1.2863617892532275,
      "grad_norm": 0.31133219599723816,
      "learning_rate": 0.00012209146892230822,
      "loss": 0.2973,
      "step": 1286
    },
    {
      "epoch": 1.2873620705823514,
      "grad_norm": 0.33981359004974365,
      "learning_rate": 0.00012198922211075778,
      "loss": 0.3245,
      "step": 1287
    },
    {
      "epoch": 1.288362351911475,
      "grad_norm": 0.3054027557373047,
      "learning_rate": 0.00012188695113702896,
      "loss": 0.3145,
      "step": 1288
    },
    {
      "epoch": 1.289362633240599,
      "grad_norm": 0.2860870659351349,
      "learning_rate": 0.00012178465611349911,
      "loss": 0.289,
      "step": 1289
    },
    {
      "epoch": 1.2903629145697226,
      "grad_norm": 0.29347696900367737,
      "learning_rate": 0.00012168233715257194,
      "loss": 0.3006,
      "step": 1290
    },
    {
      "epoch": 1.2913631958988465,
      "grad_norm": 0.29803895950317383,
      "learning_rate": 0.00012157999436667747,
      "loss": 0.3116,
      "step": 1291
    },
    {
      "epoch": 1.2923634772279704,
      "grad_norm": 0.29137060046195984,
      "learning_rate": 0.00012147762786827193,
      "loss": 0.2965,
      "step": 1292
    },
    {
      "epoch": 1.2933637585570943,
      "grad_norm": 0.2881230413913727,
      "learning_rate": 0.00012137523776983757,
      "loss": 0.3192,
      "step": 1293
    },
    {
      "epoch": 1.294364039886218,
      "grad_norm": 0.3432067930698395,
      "learning_rate": 0.00012127282418388264,
      "loss": 0.3271,
      "step": 1294
    },
    {
      "epoch": 1.2953643212153418,
      "grad_norm": 1.9920364618301392,
      "learning_rate": 0.0001211703872229411,
      "loss": 0.3115,
      "step": 1295
    },
    {
      "epoch": 1.2963646025444655,
      "grad_norm": 0.4116724729537964,
      "learning_rate": 0.00012106792699957263,
      "loss": 0.3093,
      "step": 1296
    },
    {
      "epoch": 1.2973648838735894,
      "grad_norm": 0.6236639618873596,
      "learning_rate": 0.00012096544362636255,
      "loss": 0.2905,
      "step": 1297
    },
    {
      "epoch": 1.2983651652027133,
      "grad_norm": 0.6030179262161255,
      "learning_rate": 0.00012086293721592152,
      "loss": 0.3019,
      "step": 1298
    },
    {
      "epoch": 1.2993654465318372,
      "grad_norm": 0.2736005187034607,
      "learning_rate": 0.00012076040788088554,
      "loss": 0.2762,
      "step": 1299
    },
    {
      "epoch": 1.3003657278609608,
      "grad_norm": 0.3494783639907837,
      "learning_rate": 0.00012065785573391581,
      "loss": 0.2855,
      "step": 1300
    },
    {
      "epoch": 1.3013660091900847,
      "grad_norm": 0.316986620426178,
      "learning_rate": 0.00012055528088769861,
      "loss": 0.3125,
      "step": 1301
    },
    {
      "epoch": 1.3023662905192086,
      "grad_norm": 0.2966986298561096,
      "learning_rate": 0.00012045268345494511,
      "loss": 0.2877,
      "step": 1302
    },
    {
      "epoch": 1.3033665718483323,
      "grad_norm": 0.2631240785121918,
      "learning_rate": 0.00012035006354839133,
      "loss": 0.2505,
      "step": 1303
    },
    {
      "epoch": 1.3043668531774562,
      "grad_norm": 0.4056665599346161,
      "learning_rate": 0.00012024742128079805,
      "loss": 0.2929,
      "step": 1304
    },
    {
      "epoch": 1.30536713450658,
      "grad_norm": 0.31567081809043884,
      "learning_rate": 0.00012014475676495052,
      "loss": 0.2656,
      "step": 1305
    },
    {
      "epoch": 1.306367415835704,
      "grad_norm": 0.29020872712135315,
      "learning_rate": 0.00012004207011365849,
      "loss": 0.3175,
      "step": 1306
    },
    {
      "epoch": 1.3073676971648276,
      "grad_norm": 0.6140093803405762,
      "learning_rate": 0.00011993936143975599,
      "loss": 0.2885,
      "step": 1307
    },
    {
      "epoch": 1.3083679784939515,
      "grad_norm": 0.37591552734375,
      "learning_rate": 0.00011983663085610131,
      "loss": 0.2759,
      "step": 1308
    },
    {
      "epoch": 1.3093682598230751,
      "grad_norm": 0.2984912097454071,
      "learning_rate": 0.00011973387847557676,
      "loss": 0.2815,
      "step": 1309
    },
    {
      "epoch": 1.310368541152199,
      "grad_norm": 0.2700657546520233,
      "learning_rate": 0.00011963110441108863,
      "loss": 0.2905,
      "step": 1310
    },
    {
      "epoch": 1.311368822481323,
      "grad_norm": 0.3615933060646057,
      "learning_rate": 0.000119528308775567,
      "loss": 0.315,
      "step": 1311
    },
    {
      "epoch": 1.3123691038104468,
      "grad_norm": 0.28829818964004517,
      "learning_rate": 0.00011942549168196575,
      "loss": 0.2966,
      "step": 1312
    },
    {
      "epoch": 1.3133693851395705,
      "grad_norm": 0.296802282333374,
      "learning_rate": 0.00011932265324326221,
      "loss": 0.3009,
      "step": 1313
    },
    {
      "epoch": 1.3143696664686944,
      "grad_norm": 0.329943984746933,
      "learning_rate": 0.0001192197935724573,
      "loss": 0.3199,
      "step": 1314
    },
    {
      "epoch": 1.315369947797818,
      "grad_norm": 1.6867218017578125,
      "learning_rate": 0.00011911691278257511,
      "loss": 0.3477,
      "step": 1315
    },
    {
      "epoch": 1.316370229126942,
      "grad_norm": 0.34584954380989075,
      "learning_rate": 0.0001190140109866631,
      "loss": 0.3419,
      "step": 1316
    },
    {
      "epoch": 1.3173705104560658,
      "grad_norm": 0.2954583764076233,
      "learning_rate": 0.00011891108829779165,
      "loss": 0.2557,
      "step": 1317
    },
    {
      "epoch": 1.3183707917851897,
      "grad_norm": 0.2871292233467102,
      "learning_rate": 0.00011880814482905422,
      "loss": 0.2665,
      "step": 1318
    },
    {
      "epoch": 1.3193710731143133,
      "grad_norm": 0.3024531900882721,
      "learning_rate": 0.00011870518069356709,
      "loss": 0.3116,
      "step": 1319
    },
    {
      "epoch": 1.3203713544434372,
      "grad_norm": 0.27875053882598877,
      "learning_rate": 0.0001186021960044692,
      "loss": 0.2869,
      "step": 1320
    },
    {
      "epoch": 1.3213716357725611,
      "grad_norm": 0.43009957671165466,
      "learning_rate": 0.00011849919087492211,
      "loss": 0.3012,
      "step": 1321
    },
    {
      "epoch": 1.3223719171016848,
      "grad_norm": 0.3226141333580017,
      "learning_rate": 0.00011839616541810983,
      "loss": 0.343,
      "step": 1322
    },
    {
      "epoch": 1.3233721984308087,
      "grad_norm": 0.25950679183006287,
      "learning_rate": 0.00011829311974723867,
      "loss": 0.2824,
      "step": 1323
    },
    {
      "epoch": 1.3243724797599326,
      "grad_norm": 0.2492058426141739,
      "learning_rate": 0.00011819005397553723,
      "loss": 0.2889,
      "step": 1324
    },
    {
      "epoch": 1.3253727610890562,
      "grad_norm": 0.2955023944377899,
      "learning_rate": 0.00011808696821625613,
      "loss": 0.299,
      "step": 1325
    },
    {
      "epoch": 1.32637304241818,
      "grad_norm": 0.310017853975296,
      "learning_rate": 0.000117983862582668,
      "loss": 0.3091,
      "step": 1326
    },
    {
      "epoch": 1.327373323747304,
      "grad_norm": 0.3305075764656067,
      "learning_rate": 0.00011788073718806725,
      "loss": 0.2927,
      "step": 1327
    },
    {
      "epoch": 1.3283736050764277,
      "grad_norm": 0.3950020670890808,
      "learning_rate": 0.00011777759214577006,
      "loss": 0.2979,
      "step": 1328
    },
    {
      "epoch": 1.3293738864055515,
      "grad_norm": 0.29903876781463623,
      "learning_rate": 0.00011767442756911417,
      "loss": 0.3106,
      "step": 1329
    },
    {
      "epoch": 1.3303741677346754,
      "grad_norm": 0.2557845115661621,
      "learning_rate": 0.00011757124357145881,
      "loss": 0.2606,
      "step": 1330
    },
    {
      "epoch": 1.3313744490637993,
      "grad_norm": 1.4702950716018677,
      "learning_rate": 0.00011746804026618452,
      "loss": 0.3008,
      "step": 1331
    },
    {
      "epoch": 1.332374730392923,
      "grad_norm": 0.28844839334487915,
      "learning_rate": 0.00011736481776669306,
      "loss": 0.2889,
      "step": 1332
    },
    {
      "epoch": 1.3333750117220469,
      "grad_norm": 0.31285545229911804,
      "learning_rate": 0.00011726157618640728,
      "loss": 0.2883,
      "step": 1333
    },
    {
      "epoch": 1.3343752930511705,
      "grad_norm": 0.3627071678638458,
      "learning_rate": 0.00011715831563877104,
      "loss": 0.2593,
      "step": 1334
    },
    {
      "epoch": 1.3353755743802944,
      "grad_norm": 0.3800593912601471,
      "learning_rate": 0.00011705503623724898,
      "loss": 0.3445,
      "step": 1335
    },
    {
      "epoch": 1.3363758557094183,
      "grad_norm": 0.3213560879230499,
      "learning_rate": 0.00011695173809532652,
      "loss": 0.2691,
      "step": 1336
    },
    {
      "epoch": 1.3373761370385422,
      "grad_norm": 0.3610672354698181,
      "learning_rate": 0.00011684842132650957,
      "loss": 0.29,
      "step": 1337
    },
    {
      "epoch": 1.3383764183676659,
      "grad_norm": 0.35697388648986816,
      "learning_rate": 0.00011674508604432464,
      "loss": 0.281,
      "step": 1338
    },
    {
      "epoch": 1.3393766996967897,
      "grad_norm": 0.6129295229911804,
      "learning_rate": 0.00011664173236231848,
      "loss": 0.2902,
      "step": 1339
    },
    {
      "epoch": 1.3403769810259134,
      "grad_norm": 0.3715205490589142,
      "learning_rate": 0.0001165383603940581,
      "loss": 0.306,
      "step": 1340
    },
    {
      "epoch": 1.3413772623550373,
      "grad_norm": 0.2890041470527649,
      "learning_rate": 0.00011643497025313061,
      "loss": 0.2946,
      "step": 1341
    },
    {
      "epoch": 1.3423775436841612,
      "grad_norm": 0.27793583273887634,
      "learning_rate": 0.00011633156205314309,
      "loss": 0.2899,
      "step": 1342
    },
    {
      "epoch": 1.343377825013285,
      "grad_norm": 0.3771184980869293,
      "learning_rate": 0.00011622813590772244,
      "loss": 0.2799,
      "step": 1343
    },
    {
      "epoch": 1.3443781063424087,
      "grad_norm": 0.2996048629283905,
      "learning_rate": 0.00011612469193051525,
      "loss": 0.2838,
      "step": 1344
    },
    {
      "epoch": 1.3453783876715326,
      "grad_norm": 0.2861349284648895,
      "learning_rate": 0.00011602123023518779,
      "loss": 0.2659,
      "step": 1345
    },
    {
      "epoch": 1.3463786690006565,
      "grad_norm": 0.333263635635376,
      "learning_rate": 0.00011591775093542572,
      "loss": 0.3014,
      "step": 1346
    },
    {
      "epoch": 1.3473789503297802,
      "grad_norm": 0.31445837020874023,
      "learning_rate": 0.0001158142541449341,
      "loss": 0.3604,
      "step": 1347
    },
    {
      "epoch": 1.348379231658904,
      "grad_norm": 0.26672545075416565,
      "learning_rate": 0.00011571073997743716,
      "loss": 0.3051,
      "step": 1348
    },
    {
      "epoch": 1.349379512988028,
      "grad_norm": 0.24510784447193146,
      "learning_rate": 0.0001156072085466783,
      "loss": 0.2365,
      "step": 1349
    },
    {
      "epoch": 1.3503797943171518,
      "grad_norm": 0.3197627663612366,
      "learning_rate": 0.00011550365996641979,
      "loss": 0.2895,
      "step": 1350
    },
    {
      "epoch": 1.3513800756462755,
      "grad_norm": 0.2977510392665863,
      "learning_rate": 0.00011540009435044281,
      "loss": 0.2979,
      "step": 1351
    },
    {
      "epoch": 1.3523803569753994,
      "grad_norm": 0.31227314472198486,
      "learning_rate": 0.00011529651181254723,
      "loss": 0.2892,
      "step": 1352
    },
    {
      "epoch": 1.353380638304523,
      "grad_norm": 0.29342907667160034,
      "learning_rate": 0.0001151929124665516,
      "loss": 0.2975,
      "step": 1353
    },
    {
      "epoch": 1.354380919633647,
      "grad_norm": 0.28172770142555237,
      "learning_rate": 0.00011508929642629274,
      "loss": 0.2961,
      "step": 1354
    },
    {
      "epoch": 1.3553812009627708,
      "grad_norm": 0.2900577783584595,
      "learning_rate": 0.00011498566380562601,
      "loss": 0.3011,
      "step": 1355
    },
    {
      "epoch": 1.3563814822918947,
      "grad_norm": 0.31270453333854675,
      "learning_rate": 0.0001148820147184249,
      "loss": 0.2811,
      "step": 1356
    },
    {
      "epoch": 1.3573817636210184,
      "grad_norm": 0.27891454100608826,
      "learning_rate": 0.00011477834927858104,
      "loss": 0.2659,
      "step": 1357
    },
    {
      "epoch": 1.3583820449501423,
      "grad_norm": 0.3217909038066864,
      "learning_rate": 0.00011467466760000399,
      "loss": 0.3346,
      "step": 1358
    },
    {
      "epoch": 1.359382326279266,
      "grad_norm": 0.4346146881580353,
      "learning_rate": 0.00011457096979662114,
      "loss": 0.2448,
      "step": 1359
    },
    {
      "epoch": 1.3603826076083898,
      "grad_norm": 0.34552496671676636,
      "learning_rate": 0.00011446725598237767,
      "loss": 0.3304,
      "step": 1360
    },
    {
      "epoch": 1.3613828889375137,
      "grad_norm": 0.3015654683113098,
      "learning_rate": 0.00011436352627123623,
      "loss": 0.2848,
      "step": 1361
    },
    {
      "epoch": 1.3623831702666376,
      "grad_norm": 0.508584201335907,
      "learning_rate": 0.00011425978077717709,
      "loss": 0.3131,
      "step": 1362
    },
    {
      "epoch": 1.3633834515957612,
      "grad_norm": 0.2801364064216614,
      "learning_rate": 0.00011415601961419775,
      "loss": 0.2841,
      "step": 1363
    },
    {
      "epoch": 1.3643837329248851,
      "grad_norm": 0.2569510340690613,
      "learning_rate": 0.00011405224289631295,
      "loss": 0.2668,
      "step": 1364
    },
    {
      "epoch": 1.365384014254009,
      "grad_norm": 0.38036707043647766,
      "learning_rate": 0.00011394845073755455,
      "loss": 0.2982,
      "step": 1365
    },
    {
      "epoch": 1.3663842955831327,
      "grad_norm": 0.2610377073287964,
      "learning_rate": 0.0001138446432519714,
      "loss": 0.2894,
      "step": 1366
    },
    {
      "epoch": 1.3673845769122566,
      "grad_norm": 0.316548615694046,
      "learning_rate": 0.00011374082055362909,
      "loss": 0.3744,
      "step": 1367
    },
    {
      "epoch": 1.3683848582413805,
      "grad_norm": 0.2978864908218384,
      "learning_rate": 0.00011363698275661001,
      "loss": 0.3368,
      "step": 1368
    },
    {
      "epoch": 1.3693851395705043,
      "grad_norm": 0.2853901982307434,
      "learning_rate": 0.00011353312997501313,
      "loss": 0.3343,
      "step": 1369
    },
    {
      "epoch": 1.370385420899628,
      "grad_norm": 0.2875453233718872,
      "learning_rate": 0.00011342926232295386,
      "loss": 0.3123,
      "step": 1370
    },
    {
      "epoch": 1.371385702228752,
      "grad_norm": 0.2667692303657532,
      "learning_rate": 0.00011332537991456398,
      "loss": 0.2728,
      "step": 1371
    },
    {
      "epoch": 1.3723859835578756,
      "grad_norm": 0.34361812472343445,
      "learning_rate": 0.00011322148286399147,
      "loss": 0.2871,
      "step": 1372
    },
    {
      "epoch": 1.3733862648869994,
      "grad_norm": 0.9387786984443665,
      "learning_rate": 0.0001131175712854004,
      "loss": 0.3385,
      "step": 1373
    },
    {
      "epoch": 1.3743865462161233,
      "grad_norm": 0.27623605728149414,
      "learning_rate": 0.00011301364529297079,
      "loss": 0.2924,
      "step": 1374
    },
    {
      "epoch": 1.3753868275452472,
      "grad_norm": 0.3090188503265381,
      "learning_rate": 0.0001129097050008985,
      "loss": 0.2883,
      "step": 1375
    },
    {
      "epoch": 1.3763871088743709,
      "grad_norm": 0.2835902273654938,
      "learning_rate": 0.00011280575052339514,
      "loss": 0.293,
      "step": 1376
    },
    {
      "epoch": 1.3773873902034948,
      "grad_norm": 0.4228343367576599,
      "learning_rate": 0.00011270178197468789,
      "loss": 0.321,
      "step": 1377
    },
    {
      "epoch": 1.3783876715326184,
      "grad_norm": 0.26948994398117065,
      "learning_rate": 0.00011259779946901934,
      "loss": 0.2809,
      "step": 1378
    },
    {
      "epoch": 1.3793879528617423,
      "grad_norm": 0.8430594801902771,
      "learning_rate": 0.0001124938031206475,
      "loss": 0.2542,
      "step": 1379
    },
    {
      "epoch": 1.3803882341908662,
      "grad_norm": 0.3187195062637329,
      "learning_rate": 0.00011238979304384554,
      "loss": 0.2992,
      "step": 1380
    },
    {
      "epoch": 1.38138851551999,
      "grad_norm": 0.3093203008174896,
      "learning_rate": 0.0001122857693529017,
      "loss": 0.2831,
      "step": 1381
    },
    {
      "epoch": 1.3823887968491138,
      "grad_norm": 0.42628973722457886,
      "learning_rate": 0.0001121817321621192,
      "loss": 0.3367,
      "step": 1382
    },
    {
      "epoch": 1.3833890781782376,
      "grad_norm": 0.3335116505622864,
      "learning_rate": 0.00011207768158581613,
      "loss": 0.2652,
      "step": 1383
    },
    {
      "epoch": 1.3843893595073615,
      "grad_norm": 0.3154478669166565,
      "learning_rate": 0.00011197361773832525,
      "loss": 0.3205,
      "step": 1384
    },
    {
      "epoch": 1.3853896408364852,
      "grad_norm": 0.3184400498867035,
      "learning_rate": 0.00011186954073399387,
      "loss": 0.306,
      "step": 1385
    },
    {
      "epoch": 1.386389922165609,
      "grad_norm": 0.3721691071987152,
      "learning_rate": 0.00011176545068718385,
      "loss": 0.2822,
      "step": 1386
    },
    {
      "epoch": 1.387390203494733,
      "grad_norm": 0.33508580923080444,
      "learning_rate": 0.0001116613477122713,
      "loss": 0.3019,
      "step": 1387
    },
    {
      "epoch": 1.3883904848238566,
      "grad_norm": 0.2683749496936798,
      "learning_rate": 0.00011155723192364658,
      "loss": 0.271,
      "step": 1388
    },
    {
      "epoch": 1.3893907661529805,
      "grad_norm": 0.3035573661327362,
      "learning_rate": 0.00011145310343571411,
      "loss": 0.3055,
      "step": 1389
    },
    {
      "epoch": 1.3903910474821044,
      "grad_norm": 0.3628431260585785,
      "learning_rate": 0.00011134896236289224,
      "loss": 0.2807,
      "step": 1390
    },
    {
      "epoch": 1.391391328811228,
      "grad_norm": 0.27324938774108887,
      "learning_rate": 0.0001112448088196132,
      "loss": 0.2836,
      "step": 1391
    },
    {
      "epoch": 1.392391610140352,
      "grad_norm": 0.34770703315734863,
      "learning_rate": 0.00011114064292032282,
      "loss": 0.2958,
      "step": 1392
    },
    {
      "epoch": 1.3933918914694758,
      "grad_norm": 0.5477651357650757,
      "learning_rate": 0.0001110364647794807,
      "loss": 0.3279,
      "step": 1393
    },
    {
      "epoch": 1.3943921727985997,
      "grad_norm": 0.66107577085495,
      "learning_rate": 0.00011093227451155974,
      "loss": 0.2743,
      "step": 1394
    },
    {
      "epoch": 1.3953924541277234,
      "grad_norm": 0.3737531900405884,
      "learning_rate": 0.0001108280722310462,
      "loss": 0.2875,
      "step": 1395
    },
    {
      "epoch": 1.3963927354568473,
      "grad_norm": 0.28913170099258423,
      "learning_rate": 0.0001107238580524395,
      "loss": 0.3168,
      "step": 1396
    },
    {
      "epoch": 1.397393016785971,
      "grad_norm": 0.29106587171554565,
      "learning_rate": 0.00011061963209025223,
      "loss": 0.2533,
      "step": 1397
    },
    {
      "epoch": 1.3983932981150948,
      "grad_norm": 0.2792004644870758,
      "learning_rate": 0.00011051539445900983,
      "loss": 0.2811,
      "step": 1398
    },
    {
      "epoch": 1.3993935794442187,
      "grad_norm": 0.25847047567367554,
      "learning_rate": 0.00011041114527325065,
      "loss": 0.2498,
      "step": 1399
    },
    {
      "epoch": 1.4003938607733426,
      "grad_norm": 0.7204830050468445,
      "learning_rate": 0.00011030688464752566,
      "loss": 0.2859,
      "step": 1400
    },
    {
      "epoch": 1.4013941421024663,
      "grad_norm": 0.37524187564849854,
      "learning_rate": 0.00011020261269639842,
      "loss": 0.3153,
      "step": 1401
    },
    {
      "epoch": 1.4023944234315902,
      "grad_norm": 0.283348947763443,
      "learning_rate": 0.000110098329534445,
      "loss": 0.2847,
      "step": 1402
    },
    {
      "epoch": 1.4033947047607138,
      "grad_norm": 0.2946578860282898,
      "learning_rate": 0.00010999403527625367,
      "loss": 0.2736,
      "step": 1403
    },
    {
      "epoch": 1.4043949860898377,
      "grad_norm": 0.33803313970565796,
      "learning_rate": 0.00010988973003642499,
      "loss": 0.3181,
      "step": 1404
    },
    {
      "epoch": 1.4053952674189616,
      "grad_norm": 0.28953641653060913,
      "learning_rate": 0.00010978541392957156,
      "loss": 0.3283,
      "step": 1405
    },
    {
      "epoch": 1.4063955487480855,
      "grad_norm": 0.3037947714328766,
      "learning_rate": 0.00010968108707031792,
      "loss": 0.333,
      "step": 1406
    },
    {
      "epoch": 1.4073958300772091,
      "grad_norm": 0.2842441201210022,
      "learning_rate": 0.00010957674957330042,
      "loss": 0.3033,
      "step": 1407
    },
    {
      "epoch": 1.408396111406333,
      "grad_norm": 0.302964985370636,
      "learning_rate": 0.00010947240155316707,
      "loss": 0.3096,
      "step": 1408
    },
    {
      "epoch": 1.409396392735457,
      "grad_norm": 0.3367743194103241,
      "learning_rate": 0.00010936804312457749,
      "loss": 0.2609,
      "step": 1409
    },
    {
      "epoch": 1.4103966740645806,
      "grad_norm": 0.3238588571548462,
      "learning_rate": 0.00010926367440220276,
      "loss": 0.3131,
      "step": 1410
    },
    {
      "epoch": 1.4113969553937045,
      "grad_norm": 0.3282438814640045,
      "learning_rate": 0.00010915929550072517,
      "loss": 0.2634,
      "step": 1411
    },
    {
      "epoch": 1.4123972367228284,
      "grad_norm": 0.3012658953666687,
      "learning_rate": 0.00010905490653483827,
      "loss": 0.3362,
      "step": 1412
    },
    {
      "epoch": 1.4133975180519522,
      "grad_norm": 0.27101436257362366,
      "learning_rate": 0.00010895050761924668,
      "loss": 0.2892,
      "step": 1413
    },
    {
      "epoch": 1.414397799381076,
      "grad_norm": 0.3337382972240448,
      "learning_rate": 0.00010884609886866588,
      "loss": 0.2889,
      "step": 1414
    },
    {
      "epoch": 1.4153980807101998,
      "grad_norm": 0.29577383399009705,
      "learning_rate": 0.00010874168039782227,
      "loss": 0.282,
      "step": 1415
    },
    {
      "epoch": 1.4163983620393235,
      "grad_norm": 0.367313414812088,
      "learning_rate": 0.00010863725232145286,
      "loss": 0.3714,
      "step": 1416
    },
    {
      "epoch": 1.4173986433684473,
      "grad_norm": 0.38916003704071045,
      "learning_rate": 0.00010853281475430517,
      "loss": 0.3073,
      "step": 1417
    },
    {
      "epoch": 1.4183989246975712,
      "grad_norm": 0.31319689750671387,
      "learning_rate": 0.0001084283678111372,
      "loss": 0.2846,
      "step": 1418
    },
    {
      "epoch": 1.4193992060266951,
      "grad_norm": 0.2591050863265991,
      "learning_rate": 0.00010832391160671729,
      "loss": 0.2976,
      "step": 1419
    },
    {
      "epoch": 1.4203994873558188,
      "grad_norm": 0.3168414831161499,
      "learning_rate": 0.00010821944625582392,
      "loss": 0.3234,
      "step": 1420
    },
    {
      "epoch": 1.4213997686849427,
      "grad_norm": 0.34785112738609314,
      "learning_rate": 0.00010811497187324555,
      "loss": 0.3326,
      "step": 1421
    },
    {
      "epoch": 1.4224000500140663,
      "grad_norm": 0.29518836736679077,
      "learning_rate": 0.00010801048857378071,
      "loss": 0.3594,
      "step": 1422
    },
    {
      "epoch": 1.4234003313431902,
      "grad_norm": 0.2879096567630768,
      "learning_rate": 0.00010790599647223763,
      "loss": 0.3047,
      "step": 1423
    },
    {
      "epoch": 1.424400612672314,
      "grad_norm": 0.2876695692539215,
      "learning_rate": 0.0001078014956834342,
      "loss": 0.2545,
      "step": 1424
    },
    {
      "epoch": 1.425400894001438,
      "grad_norm": 0.27527257800102234,
      "learning_rate": 0.00010769698632219794,
      "loss": 0.278,
      "step": 1425
    },
    {
      "epoch": 1.4264011753305617,
      "grad_norm": 0.527629017829895,
      "learning_rate": 0.00010759246850336572,
      "loss": 0.2634,
      "step": 1426
    },
    {
      "epoch": 1.4274014566596855,
      "grad_norm": 0.2920495569705963,
      "learning_rate": 0.0001074879423417837,
      "loss": 0.3218,
      "step": 1427
    },
    {
      "epoch": 1.4284017379888094,
      "grad_norm": 0.2818286716938019,
      "learning_rate": 0.00010738340795230721,
      "loss": 0.2748,
      "step": 1428
    },
    {
      "epoch": 1.429402019317933,
      "grad_norm": 0.32660162448883057,
      "learning_rate": 0.00010727886544980068,
      "loss": 0.3105,
      "step": 1429
    },
    {
      "epoch": 1.430402300647057,
      "grad_norm": 0.2763771116733551,
      "learning_rate": 0.00010717431494913741,
      "loss": 0.3026,
      "step": 1430
    },
    {
      "epoch": 1.4314025819761809,
      "grad_norm": 0.38270360231399536,
      "learning_rate": 0.00010706975656519946,
      "loss": 0.2411,
      "step": 1431
    },
    {
      "epoch": 1.4324028633053048,
      "grad_norm": 0.3261977434158325,
      "learning_rate": 0.00010696519041287765,
      "loss": 0.2903,
      "step": 1432
    },
    {
      "epoch": 1.4334031446344284,
      "grad_norm": 0.2844538688659668,
      "learning_rate": 0.0001068606166070712,
      "loss": 0.2827,
      "step": 1433
    },
    {
      "epoch": 1.4344034259635523,
      "grad_norm": 0.36721986532211304,
      "learning_rate": 0.00010675603526268785,
      "loss": 0.3138,
      "step": 1434
    },
    {
      "epoch": 1.435403707292676,
      "grad_norm": 0.28146329522132874,
      "learning_rate": 0.00010665144649464356,
      "loss": 0.3018,
      "step": 1435
    },
    {
      "epoch": 1.4364039886217999,
      "grad_norm": 0.317131906747818,
      "learning_rate": 0.00010654685041786249,
      "loss": 0.3707,
      "step": 1436
    },
    {
      "epoch": 1.4374042699509237,
      "grad_norm": 0.2839999198913574,
      "learning_rate": 0.00010644224714727681,
      "loss": 0.264,
      "step": 1437
    },
    {
      "epoch": 1.4384045512800476,
      "grad_norm": 0.6916788816452026,
      "learning_rate": 0.0001063376367978266,
      "loss": 0.3037,
      "step": 1438
    },
    {
      "epoch": 1.4394048326091713,
      "grad_norm": 0.3323284387588501,
      "learning_rate": 0.00010623301948445971,
      "loss": 0.3326,
      "step": 1439
    },
    {
      "epoch": 1.4404051139382952,
      "grad_norm": 0.2925950884819031,
      "learning_rate": 0.00010612839532213164,
      "loss": 0.3041,
      "step": 1440
    },
    {
      "epoch": 1.4414053952674188,
      "grad_norm": 0.701082170009613,
      "learning_rate": 0.00010602376442580544,
      "loss": 0.2841,
      "step": 1441
    },
    {
      "epoch": 1.4424056765965427,
      "grad_norm": 0.31007862091064453,
      "learning_rate": 0.00010591912691045152,
      "loss": 0.3004,
      "step": 1442
    },
    {
      "epoch": 1.4434059579256666,
      "grad_norm": 0.27657729387283325,
      "learning_rate": 0.00010581448289104758,
      "loss": 0.2705,
      "step": 1443
    },
    {
      "epoch": 1.4444062392547905,
      "grad_norm": 0.3067600131034851,
      "learning_rate": 0.00010570983248257853,
      "loss": 0.2816,
      "step": 1444
    },
    {
      "epoch": 1.4454065205839142,
      "grad_norm": 0.9092283844947815,
      "learning_rate": 0.00010560517580003617,
      "loss": 0.2222,
      "step": 1445
    },
    {
      "epoch": 1.446406801913038,
      "grad_norm": 0.37874719500541687,
      "learning_rate": 0.00010550051295841931,
      "loss": 0.2858,
      "step": 1446
    },
    {
      "epoch": 1.447407083242162,
      "grad_norm": 0.3989448845386505,
      "learning_rate": 0.00010539584407273349,
      "loss": 0.325,
      "step": 1447
    },
    {
      "epoch": 1.4484073645712856,
      "grad_norm": 0.29878175258636475,
      "learning_rate": 0.00010529116925799085,
      "loss": 0.257,
      "step": 1448
    },
    {
      "epoch": 1.4494076459004095,
      "grad_norm": 0.26630696654319763,
      "learning_rate": 0.00010518648862921012,
      "loss": 0.2614,
      "step": 1449
    },
    {
      "epoch": 1.4504079272295334,
      "grad_norm": 0.32441389560699463,
      "learning_rate": 0.00010508180230141635,
      "loss": 0.3219,
      "step": 1450
    },
    {
      "epoch": 1.451408208558657,
      "grad_norm": 0.2602307200431824,
      "learning_rate": 0.00010497711038964086,
      "loss": 0.2844,
      "step": 1451
    },
    {
      "epoch": 1.452408489887781,
      "grad_norm": 0.29980212450027466,
      "learning_rate": 0.0001048724130089212,
      "loss": 0.3014,
      "step": 1452
    },
    {
      "epoch": 1.4534087712169048,
      "grad_norm": 0.24721339344978333,
      "learning_rate": 0.00010476771027430086,
      "loss": 0.2497,
      "step": 1453
    },
    {
      "epoch": 1.4544090525460285,
      "grad_norm": 0.2951927185058594,
      "learning_rate": 0.00010466300230082911,
      "loss": 0.2894,
      "step": 1454
    },
    {
      "epoch": 1.4554093338751524,
      "grad_norm": 0.2938064932823181,
      "learning_rate": 0.00010455828920356115,
      "loss": 0.2842,
      "step": 1455
    },
    {
      "epoch": 1.4564096152042763,
      "grad_norm": 0.3401380479335785,
      "learning_rate": 0.00010445357109755771,
      "loss": 0.2888,
      "step": 1456
    },
    {
      "epoch": 1.4574098965334001,
      "grad_norm": 0.3243645131587982,
      "learning_rate": 0.00010434884809788508,
      "loss": 0.3046,
      "step": 1457
    },
    {
      "epoch": 1.4584101778625238,
      "grad_norm": 0.29797258973121643,
      "learning_rate": 0.00010424412031961484,
      "loss": 0.289,
      "step": 1458
    },
    {
      "epoch": 1.4594104591916477,
      "grad_norm": 0.31769055128097534,
      "learning_rate": 0.00010413938787782394,
      "loss": 0.2839,
      "step": 1459
    },
    {
      "epoch": 1.4604107405207714,
      "grad_norm": 0.4076133072376251,
      "learning_rate": 0.00010403465088759437,
      "loss": 0.266,
      "step": 1460
    },
    {
      "epoch": 1.4614110218498952,
      "grad_norm": 0.3065798878669739,
      "learning_rate": 0.00010392990946401313,
      "loss": 0.2699,
      "step": 1461
    },
    {
      "epoch": 1.4624113031790191,
      "grad_norm": 0.4885067641735077,
      "learning_rate": 0.00010382516372217215,
      "loss": 0.3239,
      "step": 1462
    },
    {
      "epoch": 1.463411584508143,
      "grad_norm": 0.30495181679725647,
      "learning_rate": 0.000103720413777168,
      "loss": 0.3676,
      "step": 1463
    },
    {
      "epoch": 1.4644118658372667,
      "grad_norm": 0.28350934386253357,
      "learning_rate": 0.00010361565974410192,
      "loss": 0.2648,
      "step": 1464
    },
    {
      "epoch": 1.4654121471663906,
      "grad_norm": 0.25855204463005066,
      "learning_rate": 0.00010351090173807969,
      "loss": 0.2769,
      "step": 1465
    },
    {
      "epoch": 1.4664124284955142,
      "grad_norm": 0.28751957416534424,
      "learning_rate": 0.00010340613987421137,
      "loss": 0.2875,
      "step": 1466
    },
    {
      "epoch": 1.4674127098246381,
      "grad_norm": 0.3350115716457367,
      "learning_rate": 0.00010330137426761135,
      "loss": 0.2925,
      "step": 1467
    },
    {
      "epoch": 1.468412991153762,
      "grad_norm": 0.7503853440284729,
      "learning_rate": 0.00010319660503339808,
      "loss": 0.2978,
      "step": 1468
    },
    {
      "epoch": 1.469413272482886,
      "grad_norm": 0.43129727244377136,
      "learning_rate": 0.00010309183228669397,
      "loss": 0.2938,
      "step": 1469
    },
    {
      "epoch": 1.4704135538120096,
      "grad_norm": 0.2624688148498535,
      "learning_rate": 0.00010298705614262532,
      "loss": 0.3015,
      "step": 1470
    },
    {
      "epoch": 1.4714138351411334,
      "grad_norm": 0.2895128130912781,
      "learning_rate": 0.0001028822767163222,
      "loss": 0.2864,
      "step": 1471
    },
    {
      "epoch": 1.4724141164702573,
      "grad_norm": 0.32052740454673767,
      "learning_rate": 0.00010277749412291824,
      "loss": 0.2975,
      "step": 1472
    },
    {
      "epoch": 1.473414397799381,
      "grad_norm": 0.26265427470207214,
      "learning_rate": 0.00010267270847755048,
      "loss": 0.3091,
      "step": 1473
    },
    {
      "epoch": 1.4744146791285049,
      "grad_norm": 0.3295222818851471,
      "learning_rate": 0.00010256791989535952,
      "loss": 0.2924,
      "step": 1474
    },
    {
      "epoch": 1.4754149604576288,
      "grad_norm": 0.2582465708255768,
      "learning_rate": 0.00010246312849148899,
      "loss": 0.2421,
      "step": 1475
    },
    {
      "epoch": 1.4764152417867527,
      "grad_norm": 0.2721632421016693,
      "learning_rate": 0.00010235833438108571,
      "loss": 0.2887,
      "step": 1476
    },
    {
      "epoch": 1.4774155231158763,
      "grad_norm": 0.38156336545944214,
      "learning_rate": 0.00010225353767929944,
      "loss": 0.3007,
      "step": 1477
    },
    {
      "epoch": 1.4784158044450002,
      "grad_norm": 0.29942312836647034,
      "learning_rate": 0.00010214873850128282,
      "loss": 0.2951,
      "step": 1478
    },
    {
      "epoch": 1.4794160857741239,
      "grad_norm": 0.34414979815483093,
      "learning_rate": 0.00010204393696219117,
      "loss": 0.3106,
      "step": 1479
    },
    {
      "epoch": 1.4804163671032478,
      "grad_norm": 0.30141571164131165,
      "learning_rate": 0.00010193913317718244,
      "loss": 0.2777,
      "step": 1480
    },
    {
      "epoch": 1.4814166484323716,
      "grad_norm": 0.25677481293678284,
      "learning_rate": 0.00010183432726141706,
      "loss": 0.2645,
      "step": 1481
    },
    {
      "epoch": 1.4824169297614955,
      "grad_norm": 0.26159390807151794,
      "learning_rate": 0.00010172951933005775,
      "loss": 0.2668,
      "step": 1482
    },
    {
      "epoch": 1.4834172110906192,
      "grad_norm": 0.2687196433544159,
      "learning_rate": 0.00010162470949826948,
      "loss": 0.2777,
      "step": 1483
    },
    {
      "epoch": 1.484417492419743,
      "grad_norm": 0.258765310049057,
      "learning_rate": 0.0001015198978812193,
      "loss": 0.2586,
      "step": 1484
    },
    {
      "epoch": 1.4854177737488667,
      "grad_norm": 0.3169214129447937,
      "learning_rate": 0.00010141508459407623,
      "loss": 0.3229,
      "step": 1485
    },
    {
      "epoch": 1.4864180550779906,
      "grad_norm": 0.2595394551753998,
      "learning_rate": 0.0001013102697520111,
      "loss": 0.2771,
      "step": 1486
    },
    {
      "epoch": 1.4874183364071145,
      "grad_norm": 0.27254846692085266,
      "learning_rate": 0.00010120545347019647,
      "loss": 0.2527,
      "step": 1487
    },
    {
      "epoch": 1.4884186177362384,
      "grad_norm": 0.2980557084083557,
      "learning_rate": 0.00010110063586380646,
      "loss": 0.2934,
      "step": 1488
    },
    {
      "epoch": 1.489418899065362,
      "grad_norm": 0.2934664487838745,
      "learning_rate": 0.00010099581704801673,
      "loss": 0.2869,
      "step": 1489
    },
    {
      "epoch": 1.490419180394486,
      "grad_norm": 0.4363008141517639,
      "learning_rate": 0.00010089099713800414,
      "loss": 0.3305,
      "step": 1490
    },
    {
      "epoch": 1.4914194617236098,
      "grad_norm": 0.2492322474718094,
      "learning_rate": 0.00010078617624894684,
      "loss": 0.2457,
      "step": 1491
    },
    {
      "epoch": 1.4924197430527335,
      "grad_norm": 0.32045429944992065,
      "learning_rate": 0.000100681354496024,
      "loss": 0.307,
      "step": 1492
    },
    {
      "epoch": 1.4934200243818574,
      "grad_norm": 0.28030478954315186,
      "learning_rate": 0.00010057653199441581,
      "loss": 0.2762,
      "step": 1493
    },
    {
      "epoch": 1.4944203057109813,
      "grad_norm": 0.3311432898044586,
      "learning_rate": 0.00010047170885930324,
      "loss": 0.3202,
      "step": 1494
    },
    {
      "epoch": 1.4954205870401052,
      "grad_norm": 0.30537986755371094,
      "learning_rate": 0.00010036688520586788,
      "loss": 0.2911,
      "step": 1495
    },
    {
      "epoch": 1.4964208683692288,
      "grad_norm": 0.28304117918014526,
      "learning_rate": 0.00010026206114929209,
      "loss": 0.3104,
      "step": 1496
    },
    {
      "epoch": 1.4974211496983527,
      "grad_norm": 0.3113030195236206,
      "learning_rate": 0.00010015723680475846,
      "loss": 0.3025,
      "step": 1497
    },
    {
      "epoch": 1.4984214310274764,
      "grad_norm": 0.5105164647102356,
      "learning_rate": 0.00010005241228745004,
      "loss": 0.2848,
      "step": 1498
    },
    {
      "epoch": 1.4994217123566003,
      "grad_norm": 0.286456435918808,
      "learning_rate": 9.994758771254997e-05,
      "loss": 0.2451,
      "step": 1499
    },
    {
      "epoch": 1.5004219936857242,
      "grad_norm": 0.3024538457393646,
      "learning_rate": 9.984276319524154e-05,
      "loss": 0.2762,
      "step": 1500
    },
    {
      "epoch": 1.501422275014848,
      "grad_norm": 0.27047428488731384,
      "learning_rate": 9.973793885070792e-05,
      "loss": 0.2649,
      "step": 1501
    },
    {
      "epoch": 1.5024225563439717,
      "grad_norm": 0.2758313715457916,
      "learning_rate": 9.963311479413211e-05,
      "loss": 0.3124,
      "step": 1502
    },
    {
      "epoch": 1.5034228376730956,
      "grad_norm": 0.40638795495033264,
      "learning_rate": 9.95282911406968e-05,
      "loss": 0.3119,
      "step": 1503
    },
    {
      "epoch": 1.5044231190022193,
      "grad_norm": 0.2800775468349457,
      "learning_rate": 9.942346800558421e-05,
      "loss": 0.279,
      "step": 1504
    },
    {
      "epoch": 1.5054234003313431,
      "grad_norm": 0.31468719244003296,
      "learning_rate": 9.931864550397601e-05,
      "loss": 0.3412,
      "step": 1505
    },
    {
      "epoch": 1.506423681660467,
      "grad_norm": 0.3178141713142395,
      "learning_rate": 9.921382375105318e-05,
      "loss": 0.3257,
      "step": 1506
    },
    {
      "epoch": 1.507423962989591,
      "grad_norm": 0.3203715682029724,
      "learning_rate": 9.910900286199587e-05,
      "loss": 0.2822,
      "step": 1507
    },
    {
      "epoch": 1.5084242443187148,
      "grad_norm": 0.2678610384464264,
      "learning_rate": 9.900418295198328e-05,
      "loss": 0.2808,
      "step": 1508
    },
    {
      "epoch": 1.5094245256478385,
      "grad_norm": 2.0435290336608887,
      "learning_rate": 9.889936413619356e-05,
      "loss": 0.3159,
      "step": 1509
    },
    {
      "epoch": 1.5104248069769621,
      "grad_norm": 0.2607174813747406,
      "learning_rate": 9.879454652980358e-05,
      "loss": 0.2592,
      "step": 1510
    },
    {
      "epoch": 1.511425088306086,
      "grad_norm": 0.31447839736938477,
      "learning_rate": 9.868973024798895e-05,
      "loss": 0.3182,
      "step": 1511
    },
    {
      "epoch": 1.51242536963521,
      "grad_norm": 0.2809944152832031,
      "learning_rate": 9.858491540592382e-05,
      "loss": 0.3233,
      "step": 1512
    },
    {
      "epoch": 1.5134256509643338,
      "grad_norm": 0.274942010641098,
      "learning_rate": 9.848010211878074e-05,
      "loss": 0.2468,
      "step": 1513
    },
    {
      "epoch": 1.5144259322934577,
      "grad_norm": 0.2947619557380676,
      "learning_rate": 9.837529050173052e-05,
      "loss": 0.3109,
      "step": 1514
    },
    {
      "epoch": 1.5154262136225813,
      "grad_norm": 0.32171449065208435,
      "learning_rate": 9.827048066994225e-05,
      "loss": 0.3158,
      "step": 1515
    },
    {
      "epoch": 1.516426494951705,
      "grad_norm": 0.2831699550151825,
      "learning_rate": 9.816567273858296e-05,
      "loss": 0.2959,
      "step": 1516
    },
    {
      "epoch": 1.517426776280829,
      "grad_norm": 0.25542616844177246,
      "learning_rate": 9.806086682281758e-05,
      "loss": 0.266,
      "step": 1517
    },
    {
      "epoch": 1.5184270576099528,
      "grad_norm": 0.42157018184661865,
      "learning_rate": 9.795606303780885e-05,
      "loss": 0.2572,
      "step": 1518
    },
    {
      "epoch": 1.5194273389390767,
      "grad_norm": 0.2739235758781433,
      "learning_rate": 9.785126149871722e-05,
      "loss": 0.2918,
      "step": 1519
    },
    {
      "epoch": 1.5204276202682006,
      "grad_norm": 0.301136314868927,
      "learning_rate": 9.77464623207006e-05,
      "loss": 0.2742,
      "step": 1520
    },
    {
      "epoch": 1.5214279015973242,
      "grad_norm": 0.35702282190322876,
      "learning_rate": 9.764166561891432e-05,
      "loss": 0.269,
      "step": 1521
    },
    {
      "epoch": 1.522428182926448,
      "grad_norm": 0.28325432538986206,
      "learning_rate": 9.753687150851102e-05,
      "loss": 0.3099,
      "step": 1522
    },
    {
      "epoch": 1.5234284642555718,
      "grad_norm": 0.2826971709728241,
      "learning_rate": 9.74320801046405e-05,
      "loss": 0.2791,
      "step": 1523
    },
    {
      "epoch": 1.5244287455846957,
      "grad_norm": 0.3167450428009033,
      "learning_rate": 9.732729152244953e-05,
      "loss": 0.2777,
      "step": 1524
    },
    {
      "epoch": 1.5254290269138195,
      "grad_norm": 0.32178881764411926,
      "learning_rate": 9.722250587708181e-05,
      "loss": 0.3173,
      "step": 1525
    },
    {
      "epoch": 1.5264293082429434,
      "grad_norm": 0.26520025730133057,
      "learning_rate": 9.711772328367784e-05,
      "loss": 0.2849,
      "step": 1526
    },
    {
      "epoch": 1.527429589572067,
      "grad_norm": 0.2602308392524719,
      "learning_rate": 9.70129438573747e-05,
      "loss": 0.267,
      "step": 1527
    },
    {
      "epoch": 1.528429870901191,
      "grad_norm": 0.33153626322746277,
      "learning_rate": 9.690816771330608e-05,
      "loss": 0.3206,
      "step": 1528
    },
    {
      "epoch": 1.5294301522303146,
      "grad_norm": 0.3220507502555847,
      "learning_rate": 9.680339496660192e-05,
      "loss": 0.334,
      "step": 1529
    },
    {
      "epoch": 1.5304304335594385,
      "grad_norm": 0.4581523537635803,
      "learning_rate": 9.669862573238863e-05,
      "loss": 0.3167,
      "step": 1530
    },
    {
      "epoch": 1.5314307148885624,
      "grad_norm": 0.44411808252334595,
      "learning_rate": 9.659386012578863e-05,
      "loss": 0.2611,
      "step": 1531
    },
    {
      "epoch": 1.5324309962176863,
      "grad_norm": 0.2673170268535614,
      "learning_rate": 9.648909826192033e-05,
      "loss": 0.2817,
      "step": 1532
    },
    {
      "epoch": 1.5334312775468102,
      "grad_norm": 0.2989329397678375,
      "learning_rate": 9.63843402558981e-05,
      "loss": 0.2988,
      "step": 1533
    },
    {
      "epoch": 1.5344315588759339,
      "grad_norm": 0.32206833362579346,
      "learning_rate": 9.627958622283203e-05,
      "loss": 0.3424,
      "step": 1534
    },
    {
      "epoch": 1.5354318402050575,
      "grad_norm": 0.27534934878349304,
      "learning_rate": 9.617483627782788e-05,
      "loss": 0.2715,
      "step": 1535
    },
    {
      "epoch": 1.5364321215341814,
      "grad_norm": 0.2928500175476074,
      "learning_rate": 9.607009053598689e-05,
      "loss": 0.2542,
      "step": 1536
    },
    {
      "epoch": 1.5374324028633053,
      "grad_norm": 0.3336426317691803,
      "learning_rate": 9.596534911240566e-05,
      "loss": 0.2562,
      "step": 1537
    },
    {
      "epoch": 1.5384326841924292,
      "grad_norm": 0.25877317786216736,
      "learning_rate": 9.58606121221761e-05,
      "loss": 0.2767,
      "step": 1538
    },
    {
      "epoch": 1.539432965521553,
      "grad_norm": 0.28546953201293945,
      "learning_rate": 9.57558796803852e-05,
      "loss": 0.3081,
      "step": 1539
    },
    {
      "epoch": 1.5404332468506767,
      "grad_norm": 0.7927727103233337,
      "learning_rate": 9.565115190211497e-05,
      "loss": 0.2499,
      "step": 1540
    },
    {
      "epoch": 1.5414335281798006,
      "grad_norm": 0.6823090314865112,
      "learning_rate": 9.554642890244233e-05,
      "loss": 0.3468,
      "step": 1541
    },
    {
      "epoch": 1.5424338095089243,
      "grad_norm": 0.2776161730289459,
      "learning_rate": 9.54417107964389e-05,
      "loss": 0.29,
      "step": 1542
    },
    {
      "epoch": 1.5434340908380482,
      "grad_norm": 0.31211233139038086,
      "learning_rate": 9.533699769917092e-05,
      "loss": 0.2674,
      "step": 1543
    },
    {
      "epoch": 1.544434372167172,
      "grad_norm": 0.2923029661178589,
      "learning_rate": 9.523228972569917e-05,
      "loss": 0.2955,
      "step": 1544
    },
    {
      "epoch": 1.545434653496296,
      "grad_norm": 0.39100381731987,
      "learning_rate": 9.512758699107879e-05,
      "loss": 0.2943,
      "step": 1545
    },
    {
      "epoch": 1.5464349348254196,
      "grad_norm": 0.3171774446964264,
      "learning_rate": 9.502288961035912e-05,
      "loss": 0.3086,
      "step": 1546
    },
    {
      "epoch": 1.5474352161545435,
      "grad_norm": 0.30530598759651184,
      "learning_rate": 9.491819769858366e-05,
      "loss": 0.3095,
      "step": 1547
    },
    {
      "epoch": 1.5484354974836672,
      "grad_norm": 0.29535603523254395,
      "learning_rate": 9.48135113707899e-05,
      "loss": 0.2858,
      "step": 1548
    },
    {
      "epoch": 1.549435778812791,
      "grad_norm": 0.2819267213344574,
      "learning_rate": 9.470883074200916e-05,
      "loss": 0.3193,
      "step": 1549
    },
    {
      "epoch": 1.550436060141915,
      "grad_norm": 0.3056676387786865,
      "learning_rate": 9.460415592726653e-05,
      "loss": 0.2733,
      "step": 1550
    },
    {
      "epoch": 1.5514363414710388,
      "grad_norm": 0.37359651923179626,
      "learning_rate": 9.449948704158071e-05,
      "loss": 0.2645,
      "step": 1551
    },
    {
      "epoch": 1.5524366228001627,
      "grad_norm": 0.5312703251838684,
      "learning_rate": 9.439482419996384e-05,
      "loss": 0.3384,
      "step": 1552
    },
    {
      "epoch": 1.5534369041292864,
      "grad_norm": 0.3935904800891876,
      "learning_rate": 9.42901675174215e-05,
      "loss": 0.3248,
      "step": 1553
    },
    {
      "epoch": 1.55443718545841,
      "grad_norm": 0.3585346043109894,
      "learning_rate": 9.418551710895243e-05,
      "loss": 0.285,
      "step": 1554
    },
    {
      "epoch": 1.555437466787534,
      "grad_norm": 0.3292509615421295,
      "learning_rate": 9.408087308954853e-05,
      "loss": 0.2788,
      "step": 1555
    },
    {
      "epoch": 1.5564377481166578,
      "grad_norm": 0.3465751111507416,
      "learning_rate": 9.397623557419461e-05,
      "loss": 0.2952,
      "step": 1556
    },
    {
      "epoch": 1.5574380294457817,
      "grad_norm": 0.28232482075691223,
      "learning_rate": 9.38716046778684e-05,
      "loss": 0.298,
      "step": 1557
    },
    {
      "epoch": 1.5584383107749056,
      "grad_norm": 0.3079792559146881,
      "learning_rate": 9.37669805155403e-05,
      "loss": 0.3072,
      "step": 1558
    },
    {
      "epoch": 1.5594385921040292,
      "grad_norm": 0.2697714567184448,
      "learning_rate": 9.366236320217339e-05,
      "loss": 0.2648,
      "step": 1559
    },
    {
      "epoch": 1.5604388734331531,
      "grad_norm": 0.27641308307647705,
      "learning_rate": 9.355775285272318e-05,
      "loss": 0.2903,
      "step": 1560
    },
    {
      "epoch": 1.5614391547622768,
      "grad_norm": 0.30995839834213257,
      "learning_rate": 9.34531495821375e-05,
      "loss": 0.3194,
      "step": 1561
    },
    {
      "epoch": 1.5624394360914007,
      "grad_norm": 0.2579033076763153,
      "learning_rate": 9.334855350535645e-05,
      "loss": 0.2577,
      "step": 1562
    },
    {
      "epoch": 1.5634397174205246,
      "grad_norm": 0.27374738454818726,
      "learning_rate": 9.324396473731217e-05,
      "loss": 0.2631,
      "step": 1563
    },
    {
      "epoch": 1.5644399987496485,
      "grad_norm": 0.28959032893180847,
      "learning_rate": 9.313938339292883e-05,
      "loss": 0.2694,
      "step": 1564
    },
    {
      "epoch": 1.5654402800787721,
      "grad_norm": 0.2966287434101105,
      "learning_rate": 9.303480958712239e-05,
      "loss": 0.3286,
      "step": 1565
    },
    {
      "epoch": 1.566440561407896,
      "grad_norm": 0.3589117228984833,
      "learning_rate": 9.293024343480055e-05,
      "loss": 0.3271,
      "step": 1566
    },
    {
      "epoch": 1.5674408427370197,
      "grad_norm": 0.37698066234588623,
      "learning_rate": 9.282568505086261e-05,
      "loss": 0.268,
      "step": 1567
    },
    {
      "epoch": 1.5684411240661436,
      "grad_norm": 0.33779287338256836,
      "learning_rate": 9.272113455019935e-05,
      "loss": 0.314,
      "step": 1568
    },
    {
      "epoch": 1.5694414053952674,
      "grad_norm": 0.2631189525127411,
      "learning_rate": 9.261659204769284e-05,
      "loss": 0.2768,
      "step": 1569
    },
    {
      "epoch": 1.5704416867243913,
      "grad_norm": 0.27921345829963684,
      "learning_rate": 9.251205765821636e-05,
      "loss": 0.3192,
      "step": 1570
    },
    {
      "epoch": 1.5714419680535152,
      "grad_norm": 0.2807258665561676,
      "learning_rate": 9.240753149663433e-05,
      "loss": 0.3049,
      "step": 1571
    },
    {
      "epoch": 1.5724422493826389,
      "grad_norm": 0.3822069764137268,
      "learning_rate": 9.230301367780208e-05,
      "loss": 0.2666,
      "step": 1572
    },
    {
      "epoch": 1.5734425307117625,
      "grad_norm": 0.2776041030883789,
      "learning_rate": 9.219850431656579e-05,
      "loss": 0.2825,
      "step": 1573
    },
    {
      "epoch": 1.5744428120408864,
      "grad_norm": 0.44534552097320557,
      "learning_rate": 9.209400352776237e-05,
      "loss": 0.2947,
      "step": 1574
    },
    {
      "epoch": 1.5754430933700103,
      "grad_norm": 0.29589515924453735,
      "learning_rate": 9.198951142621929e-05,
      "loss": 0.2768,
      "step": 1575
    },
    {
      "epoch": 1.5764433746991342,
      "grad_norm": 0.28000596165657043,
      "learning_rate": 9.188502812675446e-05,
      "loss": 0.2844,
      "step": 1576
    },
    {
      "epoch": 1.577443656028258,
      "grad_norm": 0.37616828083992004,
      "learning_rate": 9.178055374417612e-05,
      "loss": 0.2944,
      "step": 1577
    },
    {
      "epoch": 1.5784439373573818,
      "grad_norm": 0.2577672004699707,
      "learning_rate": 9.167608839328272e-05,
      "loss": 0.3096,
      "step": 1578
    },
    {
      "epoch": 1.5794442186865054,
      "grad_norm": 0.31146788597106934,
      "learning_rate": 9.15716321888628e-05,
      "loss": 0.2661,
      "step": 1579
    },
    {
      "epoch": 1.5804445000156293,
      "grad_norm": 0.24622854590415955,
      "learning_rate": 9.146718524569487e-05,
      "loss": 0.2561,
      "step": 1580
    },
    {
      "epoch": 1.5814447813447532,
      "grad_norm": 0.3165883719921112,
      "learning_rate": 9.136274767854716e-05,
      "loss": 0.299,
      "step": 1581
    },
    {
      "epoch": 1.582445062673877,
      "grad_norm": 0.2556282579898834,
      "learning_rate": 9.125831960217774e-05,
      "loss": 0.258,
      "step": 1582
    },
    {
      "epoch": 1.583445344003001,
      "grad_norm": 0.3023757338523865,
      "learning_rate": 9.115390113133414e-05,
      "loss": 0.2737,
      "step": 1583
    },
    {
      "epoch": 1.5844456253321246,
      "grad_norm": 0.2539224922657013,
      "learning_rate": 9.104949238075336e-05,
      "loss": 0.2675,
      "step": 1584
    },
    {
      "epoch": 1.5854459066612485,
      "grad_norm": 0.306757390499115,
      "learning_rate": 9.094509346516178e-05,
      "loss": 0.3064,
      "step": 1585
    },
    {
      "epoch": 1.5864461879903722,
      "grad_norm": 0.2849937081336975,
      "learning_rate": 9.084070449927488e-05,
      "loss": 0.2695,
      "step": 1586
    },
    {
      "epoch": 1.587446469319496,
      "grad_norm": 0.25684022903442383,
      "learning_rate": 9.07363255977973e-05,
      "loss": 0.2251,
      "step": 1587
    },
    {
      "epoch": 1.58844675064862,
      "grad_norm": 0.3035113513469696,
      "learning_rate": 9.063195687542249e-05,
      "loss": 0.2974,
      "step": 1588
    },
    {
      "epoch": 1.5894470319777438,
      "grad_norm": 0.29622411727905273,
      "learning_rate": 9.052759844683295e-05,
      "loss": 0.2625,
      "step": 1589
    },
    {
      "epoch": 1.5904473133068675,
      "grad_norm": 0.3834383487701416,
      "learning_rate": 9.042325042669961e-05,
      "loss": 0.2646,
      "step": 1590
    },
    {
      "epoch": 1.5914475946359914,
      "grad_norm": 0.2675943374633789,
      "learning_rate": 9.03189129296821e-05,
      "loss": 0.258,
      "step": 1591
    },
    {
      "epoch": 1.592447875965115,
      "grad_norm": 0.26264604926109314,
      "learning_rate": 9.021458607042845e-05,
      "loss": 0.3078,
      "step": 1592
    },
    {
      "epoch": 1.593448157294239,
      "grad_norm": 0.267078161239624,
      "learning_rate": 9.011026996357503e-05,
      "loss": 0.2517,
      "step": 1593
    },
    {
      "epoch": 1.5944484386233628,
      "grad_norm": 0.2662124037742615,
      "learning_rate": 9.000596472374637e-05,
      "loss": 0.2456,
      "step": 1594
    },
    {
      "epoch": 1.5954487199524867,
      "grad_norm": 0.3815597593784332,
      "learning_rate": 8.990167046555504e-05,
      "loss": 0.298,
      "step": 1595
    },
    {
      "epoch": 1.5964490012816106,
      "grad_norm": 0.23948106169700623,
      "learning_rate": 8.97973873036016e-05,
      "loss": 0.2429,
      "step": 1596
    },
    {
      "epoch": 1.5974492826107343,
      "grad_norm": 0.2952445447444916,
      "learning_rate": 8.969311535247438e-05,
      "loss": 0.2778,
      "step": 1597
    },
    {
      "epoch": 1.598449563939858,
      "grad_norm": 0.31881463527679443,
      "learning_rate": 8.958885472674939e-05,
      "loss": 0.3479,
      "step": 1598
    },
    {
      "epoch": 1.5994498452689818,
      "grad_norm": 0.249563530087471,
      "learning_rate": 8.948460554099018e-05,
      "loss": 0.2666,
      "step": 1599
    },
    {
      "epoch": 1.6004501265981057,
      "grad_norm": 0.3397107422351837,
      "learning_rate": 8.93803679097478e-05,
      "loss": 0.2621,
      "step": 1600
    },
    {
      "epoch": 1.6014504079272296,
      "grad_norm": 0.2910745143890381,
      "learning_rate": 8.927614194756052e-05,
      "loss": 0.3054,
      "step": 1601
    },
    {
      "epoch": 1.6024506892563535,
      "grad_norm": 0.28712934255599976,
      "learning_rate": 8.917192776895382e-05,
      "loss": 0.2884,
      "step": 1602
    },
    {
      "epoch": 1.6034509705854771,
      "grad_norm": 0.6023370027542114,
      "learning_rate": 8.906772548844026e-05,
      "loss": 0.2531,
      "step": 1603
    },
    {
      "epoch": 1.604451251914601,
      "grad_norm": 0.33640655875205994,
      "learning_rate": 8.896353522051928e-05,
      "loss": 0.2765,
      "step": 1604
    },
    {
      "epoch": 1.6054515332437247,
      "grad_norm": 0.2665030360221863,
      "learning_rate": 8.885935707967716e-05,
      "loss": 0.2706,
      "step": 1605
    },
    {
      "epoch": 1.6064518145728486,
      "grad_norm": 0.2764660716056824,
      "learning_rate": 8.875519118038684e-05,
      "loss": 0.2834,
      "step": 1606
    },
    {
      "epoch": 1.6074520959019725,
      "grad_norm": 0.22815074026584625,
      "learning_rate": 8.865103763710777e-05,
      "loss": 0.243,
      "step": 1607
    },
    {
      "epoch": 1.6084523772310964,
      "grad_norm": 0.334743469953537,
      "learning_rate": 8.854689656428591e-05,
      "loss": 0.3087,
      "step": 1608
    },
    {
      "epoch": 1.60945265856022,
      "grad_norm": 0.29739323258399963,
      "learning_rate": 8.844276807635343e-05,
      "loss": 0.2772,
      "step": 1609
    },
    {
      "epoch": 1.610452939889344,
      "grad_norm": 0.3333275318145752,
      "learning_rate": 8.833865228772871e-05,
      "loss": 0.2551,
      "step": 1610
    },
    {
      "epoch": 1.6114532212184676,
      "grad_norm": 0.5894758701324463,
      "learning_rate": 8.823454931281616e-05,
      "loss": 0.3193,
      "step": 1611
    },
    {
      "epoch": 1.6124535025475915,
      "grad_norm": 0.29251810908317566,
      "learning_rate": 8.813045926600615e-05,
      "loss": 0.2604,
      "step": 1612
    },
    {
      "epoch": 1.6134537838767153,
      "grad_norm": 0.2921118438243866,
      "learning_rate": 8.802638226167479e-05,
      "loss": 0.2829,
      "step": 1613
    },
    {
      "epoch": 1.6144540652058392,
      "grad_norm": 0.28151842951774597,
      "learning_rate": 8.792231841418391e-05,
      "loss": 0.31,
      "step": 1614
    },
    {
      "epoch": 1.6154543465349631,
      "grad_norm": 0.299330472946167,
      "learning_rate": 8.781826783788084e-05,
      "loss": 0.2849,
      "step": 1615
    },
    {
      "epoch": 1.6164546278640868,
      "grad_norm": 0.23678253591060638,
      "learning_rate": 8.771423064709837e-05,
      "loss": 0.2581,
      "step": 1616
    },
    {
      "epoch": 1.6174549091932104,
      "grad_norm": 0.32928869128227234,
      "learning_rate": 8.76102069561545e-05,
      "loss": 0.2901,
      "step": 1617
    },
    {
      "epoch": 1.6184551905223343,
      "grad_norm": 0.36277875304222107,
      "learning_rate": 8.750619687935251e-05,
      "loss": 0.2584,
      "step": 1618
    },
    {
      "epoch": 1.6194554718514582,
      "grad_norm": 0.29962241649627686,
      "learning_rate": 8.740220053098067e-05,
      "loss": 0.3018,
      "step": 1619
    },
    {
      "epoch": 1.620455753180582,
      "grad_norm": 0.2937452793121338,
      "learning_rate": 8.729821802531212e-05,
      "loss": 0.288,
      "step": 1620
    },
    {
      "epoch": 1.621456034509706,
      "grad_norm": 0.3224906027317047,
      "learning_rate": 8.719424947660487e-05,
      "loss": 0.299,
      "step": 1621
    },
    {
      "epoch": 1.6224563158388297,
      "grad_norm": 0.26013612747192383,
      "learning_rate": 8.70902949991015e-05,
      "loss": 0.2589,
      "step": 1622
    },
    {
      "epoch": 1.6234565971679535,
      "grad_norm": 0.29127228260040283,
      "learning_rate": 8.698635470702923e-05,
      "loss": 0.2747,
      "step": 1623
    },
    {
      "epoch": 1.6244568784970772,
      "grad_norm": 0.28735899925231934,
      "learning_rate": 8.688242871459963e-05,
      "loss": 0.2882,
      "step": 1624
    },
    {
      "epoch": 1.625457159826201,
      "grad_norm": 0.33507591485977173,
      "learning_rate": 8.677851713600855e-05,
      "loss": 0.3092,
      "step": 1625
    },
    {
      "epoch": 1.626457441155325,
      "grad_norm": 0.2943025231361389,
      "learning_rate": 8.667462008543603e-05,
      "loss": 0.3128,
      "step": 1626
    },
    {
      "epoch": 1.6274577224844489,
      "grad_norm": 0.3049907386302948,
      "learning_rate": 8.657073767704615e-05,
      "loss": 0.3047,
      "step": 1627
    },
    {
      "epoch": 1.6284580038135725,
      "grad_norm": 0.29979217052459717,
      "learning_rate": 8.646687002498692e-05,
      "loss": 0.2671,
      "step": 1628
    },
    {
      "epoch": 1.6294582851426964,
      "grad_norm": 0.6943816542625427,
      "learning_rate": 8.636301724339004e-05,
      "loss": 0.2798,
      "step": 1629
    },
    {
      "epoch": 1.63045856647182,
      "grad_norm": 0.2820971608161926,
      "learning_rate": 8.625917944637096e-05,
      "loss": 0.2954,
      "step": 1630
    },
    {
      "epoch": 1.631458847800944,
      "grad_norm": 0.30296966433525085,
      "learning_rate": 8.615535674802865e-05,
      "loss": 0.2968,
      "step": 1631
    },
    {
      "epoch": 1.6324591291300679,
      "grad_norm": 0.28805938363075256,
      "learning_rate": 8.605154926244543e-05,
      "loss": 0.278,
      "step": 1632
    },
    {
      "epoch": 1.6334594104591917,
      "grad_norm": 0.279803991317749,
      "learning_rate": 8.594775710368704e-05,
      "loss": 0.2931,
      "step": 1633
    },
    {
      "epoch": 1.6344596917883156,
      "grad_norm": 0.35216471552848816,
      "learning_rate": 8.584398038580226e-05,
      "loss": 0.2963,
      "step": 1634
    },
    {
      "epoch": 1.6354599731174393,
      "grad_norm": 0.34211820363998413,
      "learning_rate": 8.574021922282292e-05,
      "loss": 0.2694,
      "step": 1635
    },
    {
      "epoch": 1.636460254446563,
      "grad_norm": 0.33972254395484924,
      "learning_rate": 8.563647372876378e-05,
      "loss": 0.3306,
      "step": 1636
    },
    {
      "epoch": 1.6374605357756868,
      "grad_norm": 0.30837225914001465,
      "learning_rate": 8.553274401762237e-05,
      "loss": 0.2865,
      "step": 1637
    },
    {
      "epoch": 1.6384608171048107,
      "grad_norm": 0.2701537013053894,
      "learning_rate": 8.542903020337887e-05,
      "loss": 0.2916,
      "step": 1638
    },
    {
      "epoch": 1.6394610984339346,
      "grad_norm": 0.2673693001270294,
      "learning_rate": 8.532533239999602e-05,
      "loss": 0.2582,
      "step": 1639
    },
    {
      "epoch": 1.6404613797630585,
      "grad_norm": 0.8141276240348816,
      "learning_rate": 8.522165072141897e-05,
      "loss": 0.2958,
      "step": 1640
    },
    {
      "epoch": 1.6414616610921822,
      "grad_norm": 0.3315756022930145,
      "learning_rate": 8.511798528157512e-05,
      "loss": 0.3378,
      "step": 1641
    },
    {
      "epoch": 1.6424619424213058,
      "grad_norm": 0.41943639516830444,
      "learning_rate": 8.501433619437403e-05,
      "loss": 0.3225,
      "step": 1642
    },
    {
      "epoch": 1.6434622237504297,
      "grad_norm": 0.28424060344696045,
      "learning_rate": 8.49107035737073e-05,
      "loss": 0.294,
      "step": 1643
    },
    {
      "epoch": 1.6444625050795536,
      "grad_norm": 0.3088243007659912,
      "learning_rate": 8.480708753344846e-05,
      "loss": 0.3155,
      "step": 1644
    },
    {
      "epoch": 1.6454627864086775,
      "grad_norm": 0.2915169894695282,
      "learning_rate": 8.470348818745278e-05,
      "loss": 0.327,
      "step": 1645
    },
    {
      "epoch": 1.6464630677378014,
      "grad_norm": 0.2529820203781128,
      "learning_rate": 8.459990564955721e-05,
      "loss": 0.2483,
      "step": 1646
    },
    {
      "epoch": 1.647463349066925,
      "grad_norm": 0.2763206958770752,
      "learning_rate": 8.449634003358022e-05,
      "loss": 0.2734,
      "step": 1647
    },
    {
      "epoch": 1.648463630396049,
      "grad_norm": 0.888330340385437,
      "learning_rate": 8.43927914533217e-05,
      "loss": 0.2866,
      "step": 1648
    },
    {
      "epoch": 1.6494639117251726,
      "grad_norm": 0.3206218183040619,
      "learning_rate": 8.428926002256283e-05,
      "loss": 0.2982,
      "step": 1649
    },
    {
      "epoch": 1.6504641930542965,
      "grad_norm": 0.28334110975265503,
      "learning_rate": 8.418574585506591e-05,
      "loss": 0.2685,
      "step": 1650
    },
    {
      "epoch": 1.6514644743834204,
      "grad_norm": 0.32192686200141907,
      "learning_rate": 8.408224906457429e-05,
      "loss": 0.2753,
      "step": 1651
    },
    {
      "epoch": 1.6524647557125443,
      "grad_norm": 0.34235841035842896,
      "learning_rate": 8.397876976481224e-05,
      "loss": 0.2647,
      "step": 1652
    },
    {
      "epoch": 1.653465037041668,
      "grad_norm": 1.6083093881607056,
      "learning_rate": 8.387530806948476e-05,
      "loss": 0.2971,
      "step": 1653
    },
    {
      "epoch": 1.6544653183707918,
      "grad_norm": 0.46902158856391907,
      "learning_rate": 8.37718640922776e-05,
      "loss": 0.2874,
      "step": 1654
    },
    {
      "epoch": 1.6554655996999155,
      "grad_norm": 0.2561240494251251,
      "learning_rate": 8.366843794685695e-05,
      "loss": 0.275,
      "step": 1655
    },
    {
      "epoch": 1.6564658810290394,
      "grad_norm": 0.3820081055164337,
      "learning_rate": 8.356502974686941e-05,
      "loss": 0.2923,
      "step": 1656
    },
    {
      "epoch": 1.6574661623581632,
      "grad_norm": 0.3504351079463959,
      "learning_rate": 8.346163960594193e-05,
      "loss": 0.2925,
      "step": 1657
    },
    {
      "epoch": 1.6584664436872871,
      "grad_norm": 0.2526865005493164,
      "learning_rate": 8.335826763768156e-05,
      "loss": 0.2754,
      "step": 1658
    },
    {
      "epoch": 1.659466725016411,
      "grad_norm": 0.2689785659313202,
      "learning_rate": 8.325491395567541e-05,
      "loss": 0.2588,
      "step": 1659
    },
    {
      "epoch": 1.6604670063455347,
      "grad_norm": 0.49268078804016113,
      "learning_rate": 8.315157867349046e-05,
      "loss": 0.3008,
      "step": 1660
    },
    {
      "epoch": 1.6614672876746583,
      "grad_norm": 0.33979088068008423,
      "learning_rate": 8.30482619046735e-05,
      "loss": 0.2536,
      "step": 1661
    },
    {
      "epoch": 1.6624675690037822,
      "grad_norm": 0.32075735926628113,
      "learning_rate": 8.294496376275104e-05,
      "loss": 0.2876,
      "step": 1662
    },
    {
      "epoch": 1.6634678503329061,
      "grad_norm": 0.3104102313518524,
      "learning_rate": 8.284168436122898e-05,
      "loss": 0.3348,
      "step": 1663
    },
    {
      "epoch": 1.66446813166203,
      "grad_norm": 0.5697471499443054,
      "learning_rate": 8.273842381359273e-05,
      "loss": 0.3027,
      "step": 1664
    },
    {
      "epoch": 1.665468412991154,
      "grad_norm": 0.3330022692680359,
      "learning_rate": 8.263518223330697e-05,
      "loss": 0.3141,
      "step": 1665
    },
    {
      "epoch": 1.6664686943202776,
      "grad_norm": 0.3074687421321869,
      "learning_rate": 8.253195973381552e-05,
      "loss": 0.2647,
      "step": 1666
    },
    {
      "epoch": 1.6674689756494014,
      "grad_norm": 0.2979239225387573,
      "learning_rate": 8.242875642854121e-05,
      "loss": 0.2906,
      "step": 1667
    },
    {
      "epoch": 1.668469256978525,
      "grad_norm": 0.2534531056880951,
      "learning_rate": 8.232557243088585e-05,
      "loss": 0.2663,
      "step": 1668
    },
    {
      "epoch": 1.669469538307649,
      "grad_norm": 0.2907555103302002,
      "learning_rate": 8.222240785422996e-05,
      "loss": 0.3084,
      "step": 1669
    },
    {
      "epoch": 1.6704698196367729,
      "grad_norm": 0.25248539447784424,
      "learning_rate": 8.211926281193277e-05,
      "loss": 0.2754,
      "step": 1670
    },
    {
      "epoch": 1.6714701009658968,
      "grad_norm": 0.26373419165611267,
      "learning_rate": 8.201613741733203e-05,
      "loss": 0.2476,
      "step": 1671
    },
    {
      "epoch": 1.6724703822950204,
      "grad_norm": 0.29472628235816956,
      "learning_rate": 8.191303178374389e-05,
      "loss": 0.2883,
      "step": 1672
    },
    {
      "epoch": 1.6734706636241443,
      "grad_norm": 0.2678601145744324,
      "learning_rate": 8.180994602446279e-05,
      "loss": 0.285,
      "step": 1673
    },
    {
      "epoch": 1.674470944953268,
      "grad_norm": 0.3820549547672272,
      "learning_rate": 8.170688025276134e-05,
      "loss": 0.3546,
      "step": 1674
    },
    {
      "epoch": 1.6754712262823919,
      "grad_norm": 0.31398969888687134,
      "learning_rate": 8.160383458189022e-05,
      "loss": 0.2738,
      "step": 1675
    },
    {
      "epoch": 1.6764715076115158,
      "grad_norm": 0.27539393305778503,
      "learning_rate": 8.15008091250779e-05,
      "loss": 0.2986,
      "step": 1676
    },
    {
      "epoch": 1.6774717889406396,
      "grad_norm": 0.32477006316185,
      "learning_rate": 8.13978039955308e-05,
      "loss": 0.2795,
      "step": 1677
    },
    {
      "epoch": 1.6784720702697635,
      "grad_norm": 0.30094537138938904,
      "learning_rate": 8.12948193064329e-05,
      "loss": 0.3108,
      "step": 1678
    },
    {
      "epoch": 1.6794723515988872,
      "grad_norm": 0.2883142828941345,
      "learning_rate": 8.119185517094578e-05,
      "loss": 0.2717,
      "step": 1679
    },
    {
      "epoch": 1.6804726329280109,
      "grad_norm": 0.2946195900440216,
      "learning_rate": 8.108891170220836e-05,
      "loss": 0.2669,
      "step": 1680
    },
    {
      "epoch": 1.6814729142571347,
      "grad_norm": 0.27956122159957886,
      "learning_rate": 8.098598901333692e-05,
      "loss": 0.2664,
      "step": 1681
    },
    {
      "epoch": 1.6824731955862586,
      "grad_norm": 0.2729124426841736,
      "learning_rate": 8.088308721742491e-05,
      "loss": 0.3254,
      "step": 1682
    },
    {
      "epoch": 1.6834734769153825,
      "grad_norm": 0.40199726819992065,
      "learning_rate": 8.078020642754274e-05,
      "loss": 0.2813,
      "step": 1683
    },
    {
      "epoch": 1.6844737582445064,
      "grad_norm": 0.4167426824569702,
      "learning_rate": 8.06773467567378e-05,
      "loss": 0.266,
      "step": 1684
    },
    {
      "epoch": 1.68547403957363,
      "grad_norm": 0.2731652557849884,
      "learning_rate": 8.057450831803428e-05,
      "loss": 0.283,
      "step": 1685
    },
    {
      "epoch": 1.686474320902754,
      "grad_norm": 0.2807994782924652,
      "learning_rate": 8.047169122443302e-05,
      "loss": 0.3116,
      "step": 1686
    },
    {
      "epoch": 1.6874746022318776,
      "grad_norm": 0.28650963306427,
      "learning_rate": 8.036889558891142e-05,
      "loss": 0.2685,
      "step": 1687
    },
    {
      "epoch": 1.6884748835610015,
      "grad_norm": 0.31964048743247986,
      "learning_rate": 8.026612152442329e-05,
      "loss": 0.2995,
      "step": 1688
    },
    {
      "epoch": 1.6894751648901254,
      "grad_norm": 0.32988306879997253,
      "learning_rate": 8.016336914389874e-05,
      "loss": 0.2814,
      "step": 1689
    },
    {
      "epoch": 1.6904754462192493,
      "grad_norm": 0.35897156596183777,
      "learning_rate": 8.006063856024405e-05,
      "loss": 0.2917,
      "step": 1690
    },
    {
      "epoch": 1.691475727548373,
      "grad_norm": 0.5740484595298767,
      "learning_rate": 7.995792988634152e-05,
      "loss": 0.2684,
      "step": 1691
    },
    {
      "epoch": 1.6924760088774968,
      "grad_norm": 0.3100973963737488,
      "learning_rate": 7.985524323504948e-05,
      "loss": 0.2946,
      "step": 1692
    },
    {
      "epoch": 1.6934762902066205,
      "grad_norm": 0.26558053493499756,
      "learning_rate": 7.975257871920195e-05,
      "loss": 0.2722,
      "step": 1693
    },
    {
      "epoch": 1.6944765715357444,
      "grad_norm": 0.43157386779785156,
      "learning_rate": 7.964993645160866e-05,
      "loss": 0.2832,
      "step": 1694
    },
    {
      "epoch": 1.6954768528648683,
      "grad_norm": 0.2527351379394531,
      "learning_rate": 7.954731654505491e-05,
      "loss": 0.2784,
      "step": 1695
    },
    {
      "epoch": 1.6964771341939922,
      "grad_norm": 0.2977001368999481,
      "learning_rate": 7.944471911230142e-05,
      "loss": 0.2999,
      "step": 1696
    },
    {
      "epoch": 1.697477415523116,
      "grad_norm": 0.30368685722351074,
      "learning_rate": 7.93421442660842e-05,
      "loss": 0.3044,
      "step": 1697
    },
    {
      "epoch": 1.6984776968522397,
      "grad_norm": 0.3093944489955902,
      "learning_rate": 7.923959211911449e-05,
      "loss": 0.3118,
      "step": 1698
    },
    {
      "epoch": 1.6994779781813634,
      "grad_norm": 0.31204643845558167,
      "learning_rate": 7.91370627840785e-05,
      "loss": 0.2917,
      "step": 1699
    },
    {
      "epoch": 1.7004782595104873,
      "grad_norm": 0.4429377615451813,
      "learning_rate": 7.903455637363746e-05,
      "loss": 0.2954,
      "step": 1700
    },
    {
      "epoch": 1.7014785408396111,
      "grad_norm": 0.29816558957099915,
      "learning_rate": 7.89320730004274e-05,
      "loss": 0.2592,
      "step": 1701
    },
    {
      "epoch": 1.702478822168735,
      "grad_norm": 0.26005762815475464,
      "learning_rate": 7.882961277705895e-05,
      "loss": 0.2794,
      "step": 1702
    },
    {
      "epoch": 1.703479103497859,
      "grad_norm": 0.29064491391181946,
      "learning_rate": 7.872717581611741e-05,
      "loss": 0.2987,
      "step": 1703
    },
    {
      "epoch": 1.7044793848269826,
      "grad_norm": 0.2782478630542755,
      "learning_rate": 7.862476223016246e-05,
      "loss": 0.2838,
      "step": 1704
    },
    {
      "epoch": 1.7054796661561062,
      "grad_norm": 0.3121366500854492,
      "learning_rate": 7.852237213172812e-05,
      "loss": 0.3,
      "step": 1705
    },
    {
      "epoch": 1.7064799474852301,
      "grad_norm": 0.3515245020389557,
      "learning_rate": 7.842000563332254e-05,
      "loss": 0.283,
      "step": 1706
    },
    {
      "epoch": 1.707480228814354,
      "grad_norm": 0.2958873212337494,
      "learning_rate": 7.831766284742807e-05,
      "loss": 0.2749,
      "step": 1707
    },
    {
      "epoch": 1.708480510143478,
      "grad_norm": 0.27298587560653687,
      "learning_rate": 7.82153438865009e-05,
      "loss": 0.3086,
      "step": 1708
    },
    {
      "epoch": 1.7094807914726018,
      "grad_norm": 0.26143044233322144,
      "learning_rate": 7.811304886297104e-05,
      "loss": 0.2513,
      "step": 1709
    },
    {
      "epoch": 1.7104810728017255,
      "grad_norm": 0.4868311285972595,
      "learning_rate": 7.801077788924224e-05,
      "loss": 0.3031,
      "step": 1710
    },
    {
      "epoch": 1.7114813541308493,
      "grad_norm": 0.27339968085289,
      "learning_rate": 7.790853107769179e-05,
      "loss": 0.2682,
      "step": 1711
    },
    {
      "epoch": 1.712481635459973,
      "grad_norm": 0.23888090252876282,
      "learning_rate": 7.780630854067045e-05,
      "loss": 0.2274,
      "step": 1712
    },
    {
      "epoch": 1.713481916789097,
      "grad_norm": 0.3610135614871979,
      "learning_rate": 7.77041103905023e-05,
      "loss": 0.2816,
      "step": 1713
    },
    {
      "epoch": 1.7144821981182208,
      "grad_norm": 0.2833068370819092,
      "learning_rate": 7.760193673948461e-05,
      "loss": 0.3017,
      "step": 1714
    },
    {
      "epoch": 1.7154824794473447,
      "grad_norm": 0.2535032033920288,
      "learning_rate": 7.749978769988778e-05,
      "loss": 0.2658,
      "step": 1715
    },
    {
      "epoch": 1.7164827607764683,
      "grad_norm": 0.2733532190322876,
      "learning_rate": 7.739766338395511e-05,
      "loss": 0.2853,
      "step": 1716
    },
    {
      "epoch": 1.7174830421055922,
      "grad_norm": 0.28207990527153015,
      "learning_rate": 7.729556390390275e-05,
      "loss": 0.2933,
      "step": 1717
    },
    {
      "epoch": 1.7184833234347159,
      "grad_norm": 0.2947927415370941,
      "learning_rate": 7.719348937191957e-05,
      "loss": 0.289,
      "step": 1718
    },
    {
      "epoch": 1.7194836047638398,
      "grad_norm": 0.3000861406326294,
      "learning_rate": 7.709143990016702e-05,
      "loss": 0.2999,
      "step": 1719
    },
    {
      "epoch": 1.7204838860929637,
      "grad_norm": 0.49534645676612854,
      "learning_rate": 7.698941560077899e-05,
      "loss": 0.298,
      "step": 1720
    },
    {
      "epoch": 1.7214841674220875,
      "grad_norm": 0.2868010103702545,
      "learning_rate": 7.688741658586178e-05,
      "loss": 0.3092,
      "step": 1721
    },
    {
      "epoch": 1.7224844487512114,
      "grad_norm": 0.35814905166625977,
      "learning_rate": 7.678544296749384e-05,
      "loss": 0.285,
      "step": 1722
    },
    {
      "epoch": 1.723484730080335,
      "grad_norm": 0.4821300208568573,
      "learning_rate": 7.668349485772572e-05,
      "loss": 0.2575,
      "step": 1723
    },
    {
      "epoch": 1.7244850114094588,
      "grad_norm": 0.34388700127601624,
      "learning_rate": 7.658157236857999e-05,
      "loss": 0.313,
      "step": 1724
    },
    {
      "epoch": 1.7254852927385826,
      "grad_norm": 0.2440538853406906,
      "learning_rate": 7.6479675612051e-05,
      "loss": 0.2349,
      "step": 1725
    },
    {
      "epoch": 1.7264855740677065,
      "grad_norm": 0.28555840253829956,
      "learning_rate": 7.637780470010487e-05,
      "loss": 0.244,
      "step": 1726
    },
    {
      "epoch": 1.7274858553968304,
      "grad_norm": 0.3405389189720154,
      "learning_rate": 7.62759597446793e-05,
      "loss": 0.2841,
      "step": 1727
    },
    {
      "epoch": 1.7284861367259543,
      "grad_norm": 0.3917919397354126,
      "learning_rate": 7.617414085768351e-05,
      "loss": 0.2826,
      "step": 1728
    },
    {
      "epoch": 1.729486418055078,
      "grad_norm": 0.267138808965683,
      "learning_rate": 7.607234815099802e-05,
      "loss": 0.2654,
      "step": 1729
    },
    {
      "epoch": 1.7304866993842019,
      "grad_norm": 0.27678391337394714,
      "learning_rate": 7.597058173647458e-05,
      "loss": 0.2826,
      "step": 1730
    },
    {
      "epoch": 1.7314869807133255,
      "grad_norm": 0.3632691204547882,
      "learning_rate": 7.586884172593609e-05,
      "loss": 0.3131,
      "step": 1731
    },
    {
      "epoch": 1.7324872620424494,
      "grad_norm": 0.25983569025993347,
      "learning_rate": 7.576712823117645e-05,
      "loss": 0.2613,
      "step": 1732
    },
    {
      "epoch": 1.7334875433715733,
      "grad_norm": 0.3061681389808655,
      "learning_rate": 7.566544136396037e-05,
      "loss": 0.2747,
      "step": 1733
    },
    {
      "epoch": 1.7344878247006972,
      "grad_norm": 0.2688804268836975,
      "learning_rate": 7.556378123602334e-05,
      "loss": 0.2763,
      "step": 1734
    },
    {
      "epoch": 1.7354881060298208,
      "grad_norm": 0.2640027105808258,
      "learning_rate": 7.54621479590714e-05,
      "loss": 0.2732,
      "step": 1735
    },
    {
      "epoch": 1.7364883873589447,
      "grad_norm": 0.27694281935691833,
      "learning_rate": 7.536054164478123e-05,
      "loss": 0.298,
      "step": 1736
    },
    {
      "epoch": 1.7374886686880684,
      "grad_norm": 0.3009437918663025,
      "learning_rate": 7.525896240479976e-05,
      "loss": 0.2947,
      "step": 1737
    },
    {
      "epoch": 1.7384889500171923,
      "grad_norm": 0.2881179749965668,
      "learning_rate": 7.51574103507442e-05,
      "loss": 0.2837,
      "step": 1738
    },
    {
      "epoch": 1.7394892313463162,
      "grad_norm": 0.2530950605869293,
      "learning_rate": 7.505588559420189e-05,
      "loss": 0.2603,
      "step": 1739
    },
    {
      "epoch": 1.74048951267544,
      "grad_norm": 0.26689228415489197,
      "learning_rate": 7.495438824673016e-05,
      "loss": 0.2675,
      "step": 1740
    },
    {
      "epoch": 1.741489794004564,
      "grad_norm": 0.2643873989582062,
      "learning_rate": 7.485291841985626e-05,
      "loss": 0.2819,
      "step": 1741
    },
    {
      "epoch": 1.7424900753336876,
      "grad_norm": 0.352052241563797,
      "learning_rate": 7.475147622507717e-05,
      "loss": 0.2717,
      "step": 1742
    },
    {
      "epoch": 1.7434903566628113,
      "grad_norm": 0.2729590833187103,
      "learning_rate": 7.465006177385953e-05,
      "loss": 0.2656,
      "step": 1743
    },
    {
      "epoch": 1.7444906379919352,
      "grad_norm": 0.29946374893188477,
      "learning_rate": 7.454867517763948e-05,
      "loss": 0.2718,
      "step": 1744
    },
    {
      "epoch": 1.745490919321059,
      "grad_norm": 0.28870081901550293,
      "learning_rate": 7.444731654782253e-05,
      "loss": 0.2819,
      "step": 1745
    },
    {
      "epoch": 1.746491200650183,
      "grad_norm": 0.3682313561439514,
      "learning_rate": 7.434598599578351e-05,
      "loss": 0.2971,
      "step": 1746
    },
    {
      "epoch": 1.7474914819793068,
      "grad_norm": 0.3285645842552185,
      "learning_rate": 7.424468363286634e-05,
      "loss": 0.3029,
      "step": 1747
    },
    {
      "epoch": 1.7484917633084305,
      "grad_norm": 0.3043314218521118,
      "learning_rate": 7.414340957038406e-05,
      "loss": 0.2777,
      "step": 1748
    },
    {
      "epoch": 1.7494920446375544,
      "grad_norm": 0.30527663230895996,
      "learning_rate": 7.404216391961847e-05,
      "loss": 0.2915,
      "step": 1749
    },
    {
      "epoch": 1.750492325966678,
      "grad_norm": 0.3673238754272461,
      "learning_rate": 7.394094679182024e-05,
      "loss": 0.2792,
      "step": 1750
    },
    {
      "epoch": 1.751492607295802,
      "grad_norm": 0.4857184886932373,
      "learning_rate": 7.383975829820874e-05,
      "loss": 0.2581,
      "step": 1751
    },
    {
      "epoch": 1.7524928886249258,
      "grad_norm": 0.30140936374664307,
      "learning_rate": 7.37385985499718e-05,
      "loss": 0.2903,
      "step": 1752
    },
    {
      "epoch": 1.7534931699540497,
      "grad_norm": 0.5277259945869446,
      "learning_rate": 7.36374676582657e-05,
      "loss": 0.283,
      "step": 1753
    },
    {
      "epoch": 1.7544934512831734,
      "grad_norm": 0.3166930377483368,
      "learning_rate": 7.353636573421496e-05,
      "loss": 0.2694,
      "step": 1754
    },
    {
      "epoch": 1.7554937326122972,
      "grad_norm": 0.31407108902931213,
      "learning_rate": 7.343529288891239e-05,
      "loss": 0.275,
      "step": 1755
    },
    {
      "epoch": 1.756494013941421,
      "grad_norm": 0.2714063227176666,
      "learning_rate": 7.333424923341868e-05,
      "loss": 0.273,
      "step": 1756
    },
    {
      "epoch": 1.7574942952705448,
      "grad_norm": 0.26499733328819275,
      "learning_rate": 7.323323487876257e-05,
      "loss": 0.2412,
      "step": 1757
    },
    {
      "epoch": 1.7584945765996687,
      "grad_norm": 0.2564115524291992,
      "learning_rate": 7.313224993594057e-05,
      "loss": 0.2507,
      "step": 1758
    },
    {
      "epoch": 1.7594948579287926,
      "grad_norm": 0.37476247549057007,
      "learning_rate": 7.303129451591686e-05,
      "loss": 0.258,
      "step": 1759
    },
    {
      "epoch": 1.7604951392579165,
      "grad_norm": 0.6272186636924744,
      "learning_rate": 7.29303687296232e-05,
      "loss": 0.2533,
      "step": 1760
    },
    {
      "epoch": 1.7614954205870401,
      "grad_norm": 0.26607388257980347,
      "learning_rate": 7.282947268795877e-05,
      "loss": 0.2408,
      "step": 1761
    },
    {
      "epoch": 1.7624957019161638,
      "grad_norm": 0.2829652428627014,
      "learning_rate": 7.272860650179006e-05,
      "loss": 0.3134,
      "step": 1762
    },
    {
      "epoch": 1.7634959832452877,
      "grad_norm": 0.28897249698638916,
      "learning_rate": 7.262777028195081e-05,
      "loss": 0.296,
      "step": 1763
    },
    {
      "epoch": 1.7644962645744116,
      "grad_norm": 0.25991612672805786,
      "learning_rate": 7.252696413924174e-05,
      "loss": 0.2565,
      "step": 1764
    },
    {
      "epoch": 1.7654965459035354,
      "grad_norm": 0.31229400634765625,
      "learning_rate": 7.242618818443056e-05,
      "loss": 0.2809,
      "step": 1765
    },
    {
      "epoch": 1.7664968272326593,
      "grad_norm": 0.8654646873474121,
      "learning_rate": 7.232544252825189e-05,
      "loss": 0.3018,
      "step": 1766
    },
    {
      "epoch": 1.767497108561783,
      "grad_norm": 0.2811638116836548,
      "learning_rate": 7.222472728140695e-05,
      "loss": 0.2696,
      "step": 1767
    },
    {
      "epoch": 1.7684973898909067,
      "grad_norm": 0.2747516930103302,
      "learning_rate": 7.212404255456357e-05,
      "loss": 0.285,
      "step": 1768
    },
    {
      "epoch": 1.7694976712200305,
      "grad_norm": 0.30903688073158264,
      "learning_rate": 7.202338845835606e-05,
      "loss": 0.2696,
      "step": 1769
    },
    {
      "epoch": 1.7704979525491544,
      "grad_norm": 0.29578182101249695,
      "learning_rate": 7.192276510338507e-05,
      "loss": 0.2999,
      "step": 1770
    },
    {
      "epoch": 1.7714982338782783,
      "grad_norm": 2.09362530708313,
      "learning_rate": 7.182217260021749e-05,
      "loss": 0.2954,
      "step": 1771
    },
    {
      "epoch": 1.7724985152074022,
      "grad_norm": 0.2954619526863098,
      "learning_rate": 7.172161105938624e-05,
      "loss": 0.2599,
      "step": 1772
    },
    {
      "epoch": 1.7734987965365259,
      "grad_norm": 0.28355199098587036,
      "learning_rate": 7.162108059139032e-05,
      "loss": 0.3035,
      "step": 1773
    },
    {
      "epoch": 1.7744990778656498,
      "grad_norm": 0.2839162349700928,
      "learning_rate": 7.15205813066945e-05,
      "loss": 0.2596,
      "step": 1774
    },
    {
      "epoch": 1.7754993591947734,
      "grad_norm": 1.325466513633728,
      "learning_rate": 7.142011331572936e-05,
      "loss": 0.2472,
      "step": 1775
    },
    {
      "epoch": 1.7764996405238973,
      "grad_norm": 0.28592449426651,
      "learning_rate": 7.131967672889101e-05,
      "loss": 0.3247,
      "step": 1776
    },
    {
      "epoch": 1.7774999218530212,
      "grad_norm": 0.24921873211860657,
      "learning_rate": 7.121927165654109e-05,
      "loss": 0.2594,
      "step": 1777
    },
    {
      "epoch": 1.778500203182145,
      "grad_norm": 0.426164448261261,
      "learning_rate": 7.111889820900664e-05,
      "loss": 0.3271,
      "step": 1778
    },
    {
      "epoch": 1.7795004845112687,
      "grad_norm": 0.4734213352203369,
      "learning_rate": 7.101855649657991e-05,
      "loss": 0.2512,
      "step": 1779
    },
    {
      "epoch": 1.7805007658403926,
      "grad_norm": 0.26994478702545166,
      "learning_rate": 7.091824662951827e-05,
      "loss": 0.257,
      "step": 1780
    },
    {
      "epoch": 1.7815010471695163,
      "grad_norm": 0.2710008919239044,
      "learning_rate": 7.08179687180442e-05,
      "loss": 0.2535,
      "step": 1781
    },
    {
      "epoch": 1.7825013284986402,
      "grad_norm": 0.2933289408683777,
      "learning_rate": 7.071772287234497e-05,
      "loss": 0.281,
      "step": 1782
    },
    {
      "epoch": 1.783501609827764,
      "grad_norm": 0.25010550022125244,
      "learning_rate": 7.06175092025726e-05,
      "loss": 0.2482,
      "step": 1783
    },
    {
      "epoch": 1.784501891156888,
      "grad_norm": 0.4922553598880768,
      "learning_rate": 7.051732781884378e-05,
      "loss": 0.2954,
      "step": 1784
    },
    {
      "epoch": 1.7855021724860118,
      "grad_norm": 0.28933677077293396,
      "learning_rate": 7.041717883123977e-05,
      "loss": 0.2874,
      "step": 1785
    },
    {
      "epoch": 1.7865024538151355,
      "grad_norm": 0.4217621982097626,
      "learning_rate": 7.031706234980617e-05,
      "loss": 0.2784,
      "step": 1786
    },
    {
      "epoch": 1.7875027351442592,
      "grad_norm": 0.27882295846939087,
      "learning_rate": 7.021697848455291e-05,
      "loss": 0.2664,
      "step": 1787
    },
    {
      "epoch": 1.788503016473383,
      "grad_norm": 0.24893718957901,
      "learning_rate": 7.011692734545403e-05,
      "loss": 0.2366,
      "step": 1788
    },
    {
      "epoch": 1.789503297802507,
      "grad_norm": 0.29266056418418884,
      "learning_rate": 7.001690904244767e-05,
      "loss": 0.2876,
      "step": 1789
    },
    {
      "epoch": 1.7905035791316308,
      "grad_norm": 0.2978186309337616,
      "learning_rate": 6.991692368543584e-05,
      "loss": 0.289,
      "step": 1790
    },
    {
      "epoch": 1.7915038604607547,
      "grad_norm": 0.27409154176712036,
      "learning_rate": 6.981697138428434e-05,
      "loss": 0.2864,
      "step": 1791
    },
    {
      "epoch": 1.7925041417898784,
      "grad_norm": 0.2578350007534027,
      "learning_rate": 6.971705224882271e-05,
      "loss": 0.27,
      "step": 1792
    },
    {
      "epoch": 1.7935044231190023,
      "grad_norm": 0.25812122225761414,
      "learning_rate": 6.9617166388844e-05,
      "loss": 0.231,
      "step": 1793
    },
    {
      "epoch": 1.794504704448126,
      "grad_norm": 0.7963981032371521,
      "learning_rate": 6.951731391410468e-05,
      "loss": 0.2565,
      "step": 1794
    },
    {
      "epoch": 1.7955049857772498,
      "grad_norm": 0.3096984624862671,
      "learning_rate": 6.94174949343246e-05,
      "loss": 0.2928,
      "step": 1795
    },
    {
      "epoch": 1.7965052671063737,
      "grad_norm": 0.6516916155815125,
      "learning_rate": 6.931770955918674e-05,
      "loss": 0.2435,
      "step": 1796
    },
    {
      "epoch": 1.7975055484354976,
      "grad_norm": 0.3046545088291168,
      "learning_rate": 6.921795789833723e-05,
      "loss": 0.2879,
      "step": 1797
    },
    {
      "epoch": 1.7985058297646213,
      "grad_norm": 0.8047873973846436,
      "learning_rate": 6.911824006138503e-05,
      "loss": 0.2549,
      "step": 1798
    },
    {
      "epoch": 1.7995061110937451,
      "grad_norm": 0.327420175075531,
      "learning_rate": 6.901855615790206e-05,
      "loss": 0.3244,
      "step": 1799
    },
    {
      "epoch": 1.8005063924228688,
      "grad_norm": 0.35149964690208435,
      "learning_rate": 6.891890629742288e-05,
      "loss": 0.2335,
      "step": 1800
    },
    {
      "epoch": 1.8015066737519927,
      "grad_norm": 0.29527878761291504,
      "learning_rate": 6.88192905894447e-05,
      "loss": 0.3135,
      "step": 1801
    },
    {
      "epoch": 1.8025069550811166,
      "grad_norm": 0.30747923254966736,
      "learning_rate": 6.871970914342712e-05,
      "loss": 0.3379,
      "step": 1802
    },
    {
      "epoch": 1.8035072364102405,
      "grad_norm": 0.30348876118659973,
      "learning_rate": 6.862016206879216e-05,
      "loss": 0.2628,
      "step": 1803
    },
    {
      "epoch": 1.8045075177393644,
      "grad_norm": 0.2545010447502136,
      "learning_rate": 6.852064947492405e-05,
      "loss": 0.2666,
      "step": 1804
    },
    {
      "epoch": 1.805507799068488,
      "grad_norm": 0.2783486843109131,
      "learning_rate": 6.842117147116913e-05,
      "loss": 0.2806,
      "step": 1805
    },
    {
      "epoch": 1.8065080803976117,
      "grad_norm": 0.31390294432640076,
      "learning_rate": 6.832172816683575e-05,
      "loss": 0.2951,
      "step": 1806
    },
    {
      "epoch": 1.8075083617267356,
      "grad_norm": 0.8211212158203125,
      "learning_rate": 6.82223196711941e-05,
      "loss": 0.2397,
      "step": 1807
    },
    {
      "epoch": 1.8085086430558595,
      "grad_norm": 0.427601158618927,
      "learning_rate": 6.812294609347615e-05,
      "loss": 0.2605,
      "step": 1808
    },
    {
      "epoch": 1.8095089243849833,
      "grad_norm": 0.3007138669490814,
      "learning_rate": 6.802360754287547e-05,
      "loss": 0.3084,
      "step": 1809
    },
    {
      "epoch": 1.8105092057141072,
      "grad_norm": 0.5371061563491821,
      "learning_rate": 6.79243041285472e-05,
      "loss": 0.2558,
      "step": 1810
    },
    {
      "epoch": 1.811509487043231,
      "grad_norm": 0.27178338170051575,
      "learning_rate": 6.782503595960782e-05,
      "loss": 0.2839,
      "step": 1811
    },
    {
      "epoch": 1.8125097683723548,
      "grad_norm": 0.3919385075569153,
      "learning_rate": 6.772580314513508e-05,
      "loss": 0.2663,
      "step": 1812
    },
    {
      "epoch": 1.8135100497014784,
      "grad_norm": 0.6196280717849731,
      "learning_rate": 6.762660579416791e-05,
      "loss": 0.2841,
      "step": 1813
    },
    {
      "epoch": 1.8145103310306023,
      "grad_norm": 0.3091922700405121,
      "learning_rate": 6.752744401570625e-05,
      "loss": 0.2936,
      "step": 1814
    },
    {
      "epoch": 1.8155106123597262,
      "grad_norm": 0.31391000747680664,
      "learning_rate": 6.742831791871096e-05,
      "loss": 0.3131,
      "step": 1815
    },
    {
      "epoch": 1.81651089368885,
      "grad_norm": 0.4149378538131714,
      "learning_rate": 6.732922761210369e-05,
      "loss": 0.2958,
      "step": 1816
    },
    {
      "epoch": 1.8175111750179738,
      "grad_norm": 0.24494850635528564,
      "learning_rate": 6.723017320476679e-05,
      "loss": 0.2311,
      "step": 1817
    },
    {
      "epoch": 1.8185114563470977,
      "grad_norm": 0.2760537266731262,
      "learning_rate": 6.713115480554313e-05,
      "loss": 0.2719,
      "step": 1818
    },
    {
      "epoch": 1.8195117376762213,
      "grad_norm": 0.2778835892677307,
      "learning_rate": 6.7032172523236e-05,
      "loss": 0.3097,
      "step": 1819
    },
    {
      "epoch": 1.8205120190053452,
      "grad_norm": 0.36834806203842163,
      "learning_rate": 6.693322646660906e-05,
      "loss": 0.2687,
      "step": 1820
    },
    {
      "epoch": 1.821512300334469,
      "grad_norm": 0.2763614058494568,
      "learning_rate": 6.683431674438612e-05,
      "loss": 0.2598,
      "step": 1821
    },
    {
      "epoch": 1.822512581663593,
      "grad_norm": 0.2885820269584656,
      "learning_rate": 6.673544346525107e-05,
      "loss": 0.2673,
      "step": 1822
    },
    {
      "epoch": 1.8235128629927169,
      "grad_norm": 0.6896982192993164,
      "learning_rate": 6.663660673784777e-05,
      "loss": 0.2837,
      "step": 1823
    },
    {
      "epoch": 1.8245131443218405,
      "grad_norm": 0.2616211771965027,
      "learning_rate": 6.653780667077985e-05,
      "loss": 0.2943,
      "step": 1824
    },
    {
      "epoch": 1.8255134256509642,
      "grad_norm": 0.31288227438926697,
      "learning_rate": 6.643904337261082e-05,
      "loss": 0.2933,
      "step": 1825
    },
    {
      "epoch": 1.826513706980088,
      "grad_norm": 0.2897924780845642,
      "learning_rate": 6.634031695186362e-05,
      "loss": 0.2719,
      "step": 1826
    },
    {
      "epoch": 1.827513988309212,
      "grad_norm": 0.2964365482330322,
      "learning_rate": 6.624162751702076e-05,
      "loss": 0.3413,
      "step": 1827
    },
    {
      "epoch": 1.8285142696383359,
      "grad_norm": 0.25593850016593933,
      "learning_rate": 6.614297517652409e-05,
      "loss": 0.2741,
      "step": 1828
    },
    {
      "epoch": 1.8295145509674597,
      "grad_norm": 0.2808723747730255,
      "learning_rate": 6.604436003877464e-05,
      "loss": 0.2532,
      "step": 1829
    },
    {
      "epoch": 1.8305148322965834,
      "grad_norm": 0.5029164552688599,
      "learning_rate": 6.594578221213265e-05,
      "loss": 0.2829,
      "step": 1830
    },
    {
      "epoch": 1.831515113625707,
      "grad_norm": 0.29607683420181274,
      "learning_rate": 6.58472418049173e-05,
      "loss": 0.3148,
      "step": 1831
    },
    {
      "epoch": 1.832515394954831,
      "grad_norm": 0.25461408495903015,
      "learning_rate": 6.574873892540671e-05,
      "loss": 0.3201,
      "step": 1832
    },
    {
      "epoch": 1.8335156762839548,
      "grad_norm": 0.7538458704948425,
      "learning_rate": 6.565027368183769e-05,
      "loss": 0.2567,
      "step": 1833
    },
    {
      "epoch": 1.8345159576130787,
      "grad_norm": 0.284344345331192,
      "learning_rate": 6.555184618240577e-05,
      "loss": 0.3152,
      "step": 1834
    },
    {
      "epoch": 1.8355162389422026,
      "grad_norm": 0.26931241154670715,
      "learning_rate": 6.545345653526495e-05,
      "loss": 0.2946,
      "step": 1835
    },
    {
      "epoch": 1.8365165202713263,
      "grad_norm": 0.32232072949409485,
      "learning_rate": 6.535510484852767e-05,
      "loss": 0.3197,
      "step": 1836
    },
    {
      "epoch": 1.8375168016004502,
      "grad_norm": 0.2675272822380066,
      "learning_rate": 6.525679123026463e-05,
      "loss": 0.27,
      "step": 1837
    },
    {
      "epoch": 1.8385170829295738,
      "grad_norm": 0.27033329010009766,
      "learning_rate": 6.515851578850474e-05,
      "loss": 0.2933,
      "step": 1838
    },
    {
      "epoch": 1.8395173642586977,
      "grad_norm": 0.2994767427444458,
      "learning_rate": 6.506027863123492e-05,
      "loss": 0.2766,
      "step": 1839
    },
    {
      "epoch": 1.8405176455878216,
      "grad_norm": 0.28950828313827515,
      "learning_rate": 6.496207986640004e-05,
      "loss": 0.2955,
      "step": 1840
    },
    {
      "epoch": 1.8415179269169455,
      "grad_norm": 0.2875865697860718,
      "learning_rate": 6.48639196019028e-05,
      "loss": 0.3029,
      "step": 1841
    },
    {
      "epoch": 1.8425182082460692,
      "grad_norm": 0.2839503586292267,
      "learning_rate": 6.476579794560356e-05,
      "loss": 0.2813,
      "step": 1842
    },
    {
      "epoch": 1.843518489575193,
      "grad_norm": 0.398840993642807,
      "learning_rate": 6.46677150053203e-05,
      "loss": 0.3663,
      "step": 1843
    },
    {
      "epoch": 1.8445187709043167,
      "grad_norm": 0.2867621183395386,
      "learning_rate": 6.45696708888284e-05,
      "loss": 0.2156,
      "step": 1844
    },
    {
      "epoch": 1.8455190522334406,
      "grad_norm": 0.28682950139045715,
      "learning_rate": 6.447166570386063e-05,
      "loss": 0.2635,
      "step": 1845
    },
    {
      "epoch": 1.8465193335625645,
      "grad_norm": 0.2949700355529785,
      "learning_rate": 6.437369955810699e-05,
      "loss": 0.2828,
      "step": 1846
    },
    {
      "epoch": 1.8475196148916884,
      "grad_norm": 0.2635686695575714,
      "learning_rate": 6.42757725592145e-05,
      "loss": 0.2497,
      "step": 1847
    },
    {
      "epoch": 1.8485198962208123,
      "grad_norm": 0.28180810809135437,
      "learning_rate": 6.417788481478728e-05,
      "loss": 0.3177,
      "step": 1848
    },
    {
      "epoch": 1.849520177549936,
      "grad_norm": 0.313768595457077,
      "learning_rate": 6.40800364323862e-05,
      "loss": 0.2876,
      "step": 1849
    },
    {
      "epoch": 1.8505204588790596,
      "grad_norm": 0.2916388213634491,
      "learning_rate": 6.398222751952899e-05,
      "loss": 0.2873,
      "step": 1850
    },
    {
      "epoch": 1.8515207402081835,
      "grad_norm": 0.269709050655365,
      "learning_rate": 6.388445818368991e-05,
      "loss": 0.3022,
      "step": 1851
    },
    {
      "epoch": 1.8525210215373074,
      "grad_norm": 0.3073900043964386,
      "learning_rate": 6.378672853229981e-05,
      "loss": 0.3207,
      "step": 1852
    },
    {
      "epoch": 1.8535213028664312,
      "grad_norm": 0.3269192576408386,
      "learning_rate": 6.368903867274585e-05,
      "loss": 0.2757,
      "step": 1853
    },
    {
      "epoch": 1.8545215841955551,
      "grad_norm": 0.26112043857574463,
      "learning_rate": 6.35913887123716e-05,
      "loss": 0.2656,
      "step": 1854
    },
    {
      "epoch": 1.8555218655246788,
      "grad_norm": 0.2512447237968445,
      "learning_rate": 6.34937787584767e-05,
      "loss": 0.2349,
      "step": 1855
    },
    {
      "epoch": 1.8565221468538027,
      "grad_norm": 0.27364760637283325,
      "learning_rate": 6.339620891831678e-05,
      "loss": 0.3054,
      "step": 1856
    },
    {
      "epoch": 1.8575224281829263,
      "grad_norm": 0.273713082075119,
      "learning_rate": 6.329867929910347e-05,
      "loss": 0.2813,
      "step": 1857
    },
    {
      "epoch": 1.8585227095120502,
      "grad_norm": 0.3288629353046417,
      "learning_rate": 6.32011900080042e-05,
      "loss": 0.3296,
      "step": 1858
    },
    {
      "epoch": 1.8595229908411741,
      "grad_norm": 0.25608378648757935,
      "learning_rate": 6.310374115214204e-05,
      "loss": 0.2399,
      "step": 1859
    },
    {
      "epoch": 1.860523272170298,
      "grad_norm": 0.27778133749961853,
      "learning_rate": 6.30063328385957e-05,
      "loss": 0.2771,
      "step": 1860
    },
    {
      "epoch": 1.8615235534994217,
      "grad_norm": 0.40155231952667236,
      "learning_rate": 6.290896517439925e-05,
      "loss": 0.2566,
      "step": 1861
    },
    {
      "epoch": 1.8625238348285456,
      "grad_norm": 0.26529255509376526,
      "learning_rate": 6.281163826654218e-05,
      "loss": 0.2542,
      "step": 1862
    },
    {
      "epoch": 1.8635241161576692,
      "grad_norm": 0.2570842504501343,
      "learning_rate": 6.271435222196916e-05,
      "loss": 0.2604,
      "step": 1863
    },
    {
      "epoch": 1.864524397486793,
      "grad_norm": 0.2710798680782318,
      "learning_rate": 6.261710714757994e-05,
      "loss": 0.2728,
      "step": 1864
    },
    {
      "epoch": 1.865524678815917,
      "grad_norm": 0.2697989046573639,
      "learning_rate": 6.251990315022927e-05,
      "loss": 0.2898,
      "step": 1865
    },
    {
      "epoch": 1.8665249601450409,
      "grad_norm": 0.27998554706573486,
      "learning_rate": 6.24227403367268e-05,
      "loss": 0.2387,
      "step": 1866
    },
    {
      "epoch": 1.8675252414741648,
      "grad_norm": 0.25968292355537415,
      "learning_rate": 6.232561881383687e-05,
      "loss": 0.2524,
      "step": 1867
    },
    {
      "epoch": 1.8685255228032884,
      "grad_norm": 0.29719793796539307,
      "learning_rate": 6.222853868827839e-05,
      "loss": 0.2924,
      "step": 1868
    },
    {
      "epoch": 1.869525804132412,
      "grad_norm": 0.59858238697052,
      "learning_rate": 6.213150006672499e-05,
      "loss": 0.276,
      "step": 1869
    },
    {
      "epoch": 1.870526085461536,
      "grad_norm": 0.2908618748188019,
      "learning_rate": 6.20345030558045e-05,
      "loss": 0.2708,
      "step": 1870
    },
    {
      "epoch": 1.8715263667906599,
      "grad_norm": 0.30577144026756287,
      "learning_rate": 6.193754776209911e-05,
      "loss": 0.2669,
      "step": 1871
    },
    {
      "epoch": 1.8725266481197838,
      "grad_norm": 0.7859020829200745,
      "learning_rate": 6.184063429214515e-05,
      "loss": 0.2766,
      "step": 1872
    },
    {
      "epoch": 1.8735269294489076,
      "grad_norm": 0.2814067304134369,
      "learning_rate": 6.174376275243299e-05,
      "loss": 0.2512,
      "step": 1873
    },
    {
      "epoch": 1.8745272107780313,
      "grad_norm": 0.2789251506328583,
      "learning_rate": 6.164693324940694e-05,
      "loss": 0.3089,
      "step": 1874
    },
    {
      "epoch": 1.8755274921071552,
      "grad_norm": 0.2686850428581238,
      "learning_rate": 6.15501458894651e-05,
      "loss": 0.2614,
      "step": 1875
    },
    {
      "epoch": 1.8765277734362789,
      "grad_norm": 0.2919791340827942,
      "learning_rate": 6.145340077895929e-05,
      "loss": 0.288,
      "step": 1876
    },
    {
      "epoch": 1.8775280547654027,
      "grad_norm": 0.3091028332710266,
      "learning_rate": 6.135669802419488e-05,
      "loss": 0.3181,
      "step": 1877
    },
    {
      "epoch": 1.8785283360945266,
      "grad_norm": 0.2611595094203949,
      "learning_rate": 6.126003773143072e-05,
      "loss": 0.2771,
      "step": 1878
    },
    {
      "epoch": 1.8795286174236505,
      "grad_norm": 0.4776504933834076,
      "learning_rate": 6.116342000687896e-05,
      "loss": 0.2627,
      "step": 1879
    },
    {
      "epoch": 1.8805288987527742,
      "grad_norm": 0.32856810092926025,
      "learning_rate": 6.106684495670506e-05,
      "loss": 0.3033,
      "step": 1880
    },
    {
      "epoch": 1.881529180081898,
      "grad_norm": 0.3717559576034546,
      "learning_rate": 6.097031268702746e-05,
      "loss": 0.3034,
      "step": 1881
    },
    {
      "epoch": 1.8825294614110217,
      "grad_norm": 0.2626720666885376,
      "learning_rate": 6.087382330391774e-05,
      "loss": 0.2605,
      "step": 1882
    },
    {
      "epoch": 1.8835297427401456,
      "grad_norm": 0.2744326591491699,
      "learning_rate": 6.077737691340023e-05,
      "loss": 0.2892,
      "step": 1883
    },
    {
      "epoch": 1.8845300240692695,
      "grad_norm": 0.2658277750015259,
      "learning_rate": 6.0680973621452105e-05,
      "loss": 0.2619,
      "step": 1884
    },
    {
      "epoch": 1.8855303053983934,
      "grad_norm": 0.26971256732940674,
      "learning_rate": 6.0584613534003144e-05,
      "loss": 0.2637,
      "step": 1885
    },
    {
      "epoch": 1.8865305867275173,
      "grad_norm": 0.279358446598053,
      "learning_rate": 6.0488296756935636e-05,
      "loss": 0.2949,
      "step": 1886
    },
    {
      "epoch": 1.887530868056641,
      "grad_norm": 0.2574153542518616,
      "learning_rate": 6.039202339608432e-05,
      "loss": 0.2808,
      "step": 1887
    },
    {
      "epoch": 1.8885311493857646,
      "grad_norm": 0.276671439409256,
      "learning_rate": 6.0295793557236203e-05,
      "loss": 0.2653,
      "step": 1888
    },
    {
      "epoch": 1.8895314307148885,
      "grad_norm": 0.35512250661849976,
      "learning_rate": 6.019960734613047e-05,
      "loss": 0.3065,
      "step": 1889
    },
    {
      "epoch": 1.8905317120440124,
      "grad_norm": 0.2662924826145172,
      "learning_rate": 6.010346486845837e-05,
      "loss": 0.2649,
      "step": 1890
    },
    {
      "epoch": 1.8915319933731363,
      "grad_norm": 0.3023977279663086,
      "learning_rate": 6.0007366229863117e-05,
      "loss": 0.3069,
      "step": 1891
    },
    {
      "epoch": 1.8925322747022602,
      "grad_norm": 0.27044034004211426,
      "learning_rate": 5.991131153593971e-05,
      "loss": 0.2746,
      "step": 1892
    },
    {
      "epoch": 1.8935325560313838,
      "grad_norm": 0.31923601031303406,
      "learning_rate": 5.981530089223489e-05,
      "loss": 0.2913,
      "step": 1893
    },
    {
      "epoch": 1.8945328373605075,
      "grad_norm": 0.2557383179664612,
      "learning_rate": 5.971933440424703e-05,
      "loss": 0.2547,
      "step": 1894
    },
    {
      "epoch": 1.8955331186896314,
      "grad_norm": 0.32849571108818054,
      "learning_rate": 5.9623412177425886e-05,
      "loss": 0.3051,
      "step": 1895
    },
    {
      "epoch": 1.8965334000187553,
      "grad_norm": 0.30909809470176697,
      "learning_rate": 5.952753431717268e-05,
      "loss": 0.2908,
      "step": 1896
    },
    {
      "epoch": 1.8975336813478791,
      "grad_norm": 0.2645419239997864,
      "learning_rate": 5.9431700928839805e-05,
      "loss": 0.2859,
      "step": 1897
    },
    {
      "epoch": 1.898533962677003,
      "grad_norm": 0.29352524876594543,
      "learning_rate": 5.933591211773082e-05,
      "loss": 0.307,
      "step": 1898
    },
    {
      "epoch": 1.8995342440061267,
      "grad_norm": 0.2947612404823303,
      "learning_rate": 5.924016798910037e-05,
      "loss": 0.2651,
      "step": 1899
    },
    {
      "epoch": 1.9005345253352506,
      "grad_norm": 0.2675304412841797,
      "learning_rate": 5.914446864815388e-05,
      "loss": 0.2856,
      "step": 1900
    },
    {
      "epoch": 1.9015348066643742,
      "grad_norm": 0.23902463912963867,
      "learning_rate": 5.9048814200047675e-05,
      "loss": 0.2488,
      "step": 1901
    },
    {
      "epoch": 1.9025350879934981,
      "grad_norm": 0.2610028386116028,
      "learning_rate": 5.895320474988864e-05,
      "loss": 0.2692,
      "step": 1902
    },
    {
      "epoch": 1.903535369322622,
      "grad_norm": 0.31277430057525635,
      "learning_rate": 5.885764040273426e-05,
      "loss": 0.3184,
      "step": 1903
    },
    {
      "epoch": 1.904535650651746,
      "grad_norm": 0.26557254791259766,
      "learning_rate": 5.876212126359251e-05,
      "loss": 0.2576,
      "step": 1904
    },
    {
      "epoch": 1.9055359319808696,
      "grad_norm": 0.2657122313976288,
      "learning_rate": 5.866664743742162e-05,
      "loss": 0.2662,
      "step": 1905
    },
    {
      "epoch": 1.9065362133099935,
      "grad_norm": 0.3060644268989563,
      "learning_rate": 5.857121902913008e-05,
      "loss": 0.2597,
      "step": 1906
    },
    {
      "epoch": 1.9075364946391171,
      "grad_norm": 0.3291439712047577,
      "learning_rate": 5.8475836143576433e-05,
      "loss": 0.2877,
      "step": 1907
    },
    {
      "epoch": 1.908536775968241,
      "grad_norm": 0.287523090839386,
      "learning_rate": 5.838049888556925e-05,
      "loss": 0.3151,
      "step": 1908
    },
    {
      "epoch": 1.909537057297365,
      "grad_norm": 0.45122969150543213,
      "learning_rate": 5.8285207359866936e-05,
      "loss": 0.2563,
      "step": 1909
    },
    {
      "epoch": 1.9105373386264888,
      "grad_norm": 3.276195764541626,
      "learning_rate": 5.8189961671177574e-05,
      "loss": 0.2494,
      "step": 1910
    },
    {
      "epoch": 1.9115376199556127,
      "grad_norm": 0.29331672191619873,
      "learning_rate": 5.809476192415905e-05,
      "loss": 0.2635,
      "step": 1911
    },
    {
      "epoch": 1.9125379012847363,
      "grad_norm": 0.31475207209587097,
      "learning_rate": 5.7999608223418534e-05,
      "loss": 0.2896,
      "step": 1912
    },
    {
      "epoch": 1.91353818261386,
      "grad_norm": 0.3246888220310211,
      "learning_rate": 5.790450067351291e-05,
      "loss": 0.3094,
      "step": 1913
    },
    {
      "epoch": 1.9145384639429839,
      "grad_norm": 0.2736388146877289,
      "learning_rate": 5.780943937894805e-05,
      "loss": 0.2846,
      "step": 1914
    },
    {
      "epoch": 1.9155387452721078,
      "grad_norm": 0.2683006525039673,
      "learning_rate": 5.771442444417918e-05,
      "loss": 0.2692,
      "step": 1915
    },
    {
      "epoch": 1.9165390266012317,
      "grad_norm": 0.2680269777774811,
      "learning_rate": 5.761945597361054e-05,
      "loss": 0.2582,
      "step": 1916
    },
    {
      "epoch": 1.9175393079303555,
      "grad_norm": 0.32143768668174744,
      "learning_rate": 5.752453407159522e-05,
      "loss": 0.2822,
      "step": 1917
    },
    {
      "epoch": 1.9185395892594792,
      "grad_norm": 0.2571329176425934,
      "learning_rate": 5.742965884243532e-05,
      "loss": 0.2448,
      "step": 1918
    },
    {
      "epoch": 1.919539870588603,
      "grad_norm": 0.6229331493377686,
      "learning_rate": 5.733483039038149e-05,
      "loss": 0.2702,
      "step": 1919
    },
    {
      "epoch": 1.9205401519177268,
      "grad_norm": 0.3583178222179413,
      "learning_rate": 5.724004881963311e-05,
      "loss": 0.2633,
      "step": 1920
    },
    {
      "epoch": 1.9215404332468506,
      "grad_norm": 0.2952008545398712,
      "learning_rate": 5.714531423433791e-05,
      "loss": 0.2883,
      "step": 1921
    },
    {
      "epoch": 1.9225407145759745,
      "grad_norm": 0.27407681941986084,
      "learning_rate": 5.705062673859216e-05,
      "loss": 0.2865,
      "step": 1922
    },
    {
      "epoch": 1.9235409959050984,
      "grad_norm": 0.4782307744026184,
      "learning_rate": 5.69559864364402e-05,
      "loss": 0.2694,
      "step": 1923
    },
    {
      "epoch": 1.924541277234222,
      "grad_norm": 0.3665333390235901,
      "learning_rate": 5.6861393431874675e-05,
      "loss": 0.2843,
      "step": 1924
    },
    {
      "epoch": 1.925541558563346,
      "grad_norm": 0.2731457054615021,
      "learning_rate": 5.676684782883615e-05,
      "loss": 0.2626,
      "step": 1925
    },
    {
      "epoch": 1.9265418398924696,
      "grad_norm": 0.36634135246276855,
      "learning_rate": 5.667234973121317e-05,
      "loss": 0.2809,
      "step": 1926
    },
    {
      "epoch": 1.9275421212215935,
      "grad_norm": 0.3025986850261688,
      "learning_rate": 5.6577899242842025e-05,
      "loss": 0.3167,
      "step": 1927
    },
    {
      "epoch": 1.9285424025507174,
      "grad_norm": 0.26379942893981934,
      "learning_rate": 5.648349646750673e-05,
      "loss": 0.2793,
      "step": 1928
    },
    {
      "epoch": 1.9295426838798413,
      "grad_norm": 0.26326414942741394,
      "learning_rate": 5.6389141508938903e-05,
      "loss": 0.277,
      "step": 1929
    },
    {
      "epoch": 1.9305429652089652,
      "grad_norm": 0.3265363276004791,
      "learning_rate": 5.629483447081751e-05,
      "loss": 0.2901,
      "step": 1930
    },
    {
      "epoch": 1.9315432465380888,
      "grad_norm": 0.4215918183326721,
      "learning_rate": 5.620057545676901e-05,
      "loss": 0.2335,
      "step": 1931
    },
    {
      "epoch": 1.9325435278672125,
      "grad_norm": 0.27217063307762146,
      "learning_rate": 5.610636457036693e-05,
      "loss": 0.2704,
      "step": 1932
    },
    {
      "epoch": 1.9335438091963364,
      "grad_norm": 0.27226927876472473,
      "learning_rate": 5.601220191513208e-05,
      "loss": 0.2593,
      "step": 1933
    },
    {
      "epoch": 1.9345440905254603,
      "grad_norm": 0.23594869673252106,
      "learning_rate": 5.591808759453214e-05,
      "loss": 0.2368,
      "step": 1934
    },
    {
      "epoch": 1.9355443718545842,
      "grad_norm": 0.28655052185058594,
      "learning_rate": 5.5824021711981686e-05,
      "loss": 0.2707,
      "step": 1935
    },
    {
      "epoch": 1.936544653183708,
      "grad_norm": 0.2927952706813812,
      "learning_rate": 5.573000437084221e-05,
      "loss": 0.3027,
      "step": 1936
    },
    {
      "epoch": 1.9375449345128317,
      "grad_norm": 1.0928081274032593,
      "learning_rate": 5.563603567442168e-05,
      "loss": 0.3019,
      "step": 1937
    },
    {
      "epoch": 1.9385452158419556,
      "grad_norm": 0.41158583760261536,
      "learning_rate": 5.554211572597477e-05,
      "loss": 0.2505,
      "step": 1938
    },
    {
      "epoch": 1.9395454971710793,
      "grad_norm": 0.24977841973304749,
      "learning_rate": 5.544824462870244e-05,
      "loss": 0.2384,
      "step": 1939
    },
    {
      "epoch": 1.9405457785002032,
      "grad_norm": 0.2562898099422455,
      "learning_rate": 5.5354422485752125e-05,
      "loss": 0.2634,
      "step": 1940
    },
    {
      "epoch": 1.941546059829327,
      "grad_norm": 0.29187583923339844,
      "learning_rate": 5.5260649400217326e-05,
      "loss": 0.2737,
      "step": 1941
    },
    {
      "epoch": 1.942546341158451,
      "grad_norm": 0.2478390336036682,
      "learning_rate": 5.5166925475137735e-05,
      "loss": 0.2525,
      "step": 1942
    },
    {
      "epoch": 1.9435466224875746,
      "grad_norm": 0.27928054332733154,
      "learning_rate": 5.507325081349903e-05,
      "loss": 0.3009,
      "step": 1943
    },
    {
      "epoch": 1.9445469038166985,
      "grad_norm": 0.2793390452861786,
      "learning_rate": 5.497962551823266e-05,
      "loss": 0.2888,
      "step": 1944
    },
    {
      "epoch": 1.9455471851458221,
      "grad_norm": 0.2575443685054779,
      "learning_rate": 5.488604969221597e-05,
      "loss": 0.2458,
      "step": 1945
    },
    {
      "epoch": 1.946547466474946,
      "grad_norm": 0.49242281913757324,
      "learning_rate": 5.479252343827178e-05,
      "loss": 0.2687,
      "step": 1946
    },
    {
      "epoch": 1.94754774780407,
      "grad_norm": 0.29565486311912537,
      "learning_rate": 5.469904685916861e-05,
      "loss": 0.2615,
      "step": 1947
    },
    {
      "epoch": 1.9485480291331938,
      "grad_norm": 0.2529813051223755,
      "learning_rate": 5.460562005762024e-05,
      "loss": 0.2426,
      "step": 1948
    },
    {
      "epoch": 1.9495483104623177,
      "grad_norm": 0.26535025238990784,
      "learning_rate": 5.4512243136285915e-05,
      "loss": 0.2484,
      "step": 1949
    },
    {
      "epoch": 1.9505485917914414,
      "grad_norm": 0.3218567967414856,
      "learning_rate": 5.441891619776987e-05,
      "loss": 0.2814,
      "step": 1950
    },
    {
      "epoch": 1.951548873120565,
      "grad_norm": 0.3390984535217285,
      "learning_rate": 5.432563934462166e-05,
      "loss": 0.3075,
      "step": 1951
    },
    {
      "epoch": 1.952549154449689,
      "grad_norm": 0.30252760648727417,
      "learning_rate": 5.423241267933557e-05,
      "loss": 0.2679,
      "step": 1952
    },
    {
      "epoch": 1.9535494357788128,
      "grad_norm": 0.3998994529247284,
      "learning_rate": 5.4139236304350935e-05,
      "loss": 0.2781,
      "step": 1953
    },
    {
      "epoch": 1.9545497171079367,
      "grad_norm": 0.2822798788547516,
      "learning_rate": 5.404611032205169e-05,
      "loss": 0.2739,
      "step": 1954
    },
    {
      "epoch": 1.9555499984370606,
      "grad_norm": 0.27554377913475037,
      "learning_rate": 5.3953034834766416e-05,
      "loss": 0.2467,
      "step": 1955
    },
    {
      "epoch": 1.9565502797661842,
      "grad_norm": 0.3258563280105591,
      "learning_rate": 5.386000994476832e-05,
      "loss": 0.2509,
      "step": 1956
    },
    {
      "epoch": 1.957550561095308,
      "grad_norm": 1.2052422761917114,
      "learning_rate": 5.376703575427481e-05,
      "loss": 0.2732,
      "step": 1957
    },
    {
      "epoch": 1.9585508424244318,
      "grad_norm": 0.33804357051849365,
      "learning_rate": 5.367411236544786e-05,
      "loss": 0.2685,
      "step": 1958
    },
    {
      "epoch": 1.9595511237535557,
      "grad_norm": 0.2446807324886322,
      "learning_rate": 5.3581239880393375e-05,
      "loss": 0.2483,
      "step": 1959
    },
    {
      "epoch": 1.9605514050826796,
      "grad_norm": 0.28413617610931396,
      "learning_rate": 5.3488418401161475e-05,
      "loss": 0.2934,
      "step": 1960
    },
    {
      "epoch": 1.9615516864118034,
      "grad_norm": 0.26250582933425903,
      "learning_rate": 5.339564802974615e-05,
      "loss": 0.2578,
      "step": 1961
    },
    {
      "epoch": 1.962551967740927,
      "grad_norm": 0.3940134346485138,
      "learning_rate": 5.33029288680852e-05,
      "loss": 0.2504,
      "step": 1962
    },
    {
      "epoch": 1.963552249070051,
      "grad_norm": 0.2663330137729645,
      "learning_rate": 5.321026101806032e-05,
      "loss": 0.276,
      "step": 1963
    },
    {
      "epoch": 1.9645525303991747,
      "grad_norm": 0.27061977982521057,
      "learning_rate": 5.311764458149664e-05,
      "loss": 0.2846,
      "step": 1964
    },
    {
      "epoch": 1.9655528117282985,
      "grad_norm": 0.3375605046749115,
      "learning_rate": 5.302507966016295e-05,
      "loss": 0.2986,
      "step": 1965
    },
    {
      "epoch": 1.9665530930574224,
      "grad_norm": 0.2750922441482544,
      "learning_rate": 5.293256635577126e-05,
      "loss": 0.2538,
      "step": 1966
    },
    {
      "epoch": 1.9675533743865463,
      "grad_norm": 0.41357871890068054,
      "learning_rate": 5.284010476997705e-05,
      "loss": 0.2924,
      "step": 1967
    },
    {
      "epoch": 1.96855365571567,
      "grad_norm": 0.3273179531097412,
      "learning_rate": 5.274769500437882e-05,
      "loss": 0.285,
      "step": 1968
    },
    {
      "epoch": 1.9695539370447939,
      "grad_norm": 0.30573540925979614,
      "learning_rate": 5.265533716051825e-05,
      "loss": 0.2393,
      "step": 1969
    },
    {
      "epoch": 1.9705542183739175,
      "grad_norm": 0.3043791353702545,
      "learning_rate": 5.256303133987982e-05,
      "loss": 0.2965,
      "step": 1970
    },
    {
      "epoch": 1.9715544997030414,
      "grad_norm": 0.32803821563720703,
      "learning_rate": 5.247077764389099e-05,
      "loss": 0.3519,
      "step": 1971
    },
    {
      "epoch": 1.9725547810321653,
      "grad_norm": 0.2959039807319641,
      "learning_rate": 5.2378576173921934e-05,
      "loss": 0.3002,
      "step": 1972
    },
    {
      "epoch": 1.9735550623612892,
      "grad_norm": 0.2648478150367737,
      "learning_rate": 5.22864270312853e-05,
      "loss": 0.258,
      "step": 1973
    },
    {
      "epoch": 1.974555343690413,
      "grad_norm": 0.34027647972106934,
      "learning_rate": 5.219433031723641e-05,
      "loss": 0.2938,
      "step": 1974
    },
    {
      "epoch": 1.9755556250195367,
      "grad_norm": 0.33363237977027893,
      "learning_rate": 5.210228613297281e-05,
      "loss": 0.2554,
      "step": 1975
    },
    {
      "epoch": 1.9765559063486604,
      "grad_norm": 0.2849772274494171,
      "learning_rate": 5.201029457963451e-05,
      "loss": 0.2945,
      "step": 1976
    },
    {
      "epoch": 1.9775561876777843,
      "grad_norm": 0.5678755640983582,
      "learning_rate": 5.191835575830352e-05,
      "loss": 0.2536,
      "step": 1977
    },
    {
      "epoch": 1.9785564690069082,
      "grad_norm": 0.2611503303050995,
      "learning_rate": 5.1826469770004026e-05,
      "loss": 0.2542,
      "step": 1978
    },
    {
      "epoch": 1.979556750336032,
      "grad_norm": 0.30301058292388916,
      "learning_rate": 5.1734636715702043e-05,
      "loss": 0.2731,
      "step": 1979
    },
    {
      "epoch": 1.980557031665156,
      "grad_norm": 0.23370805382728577,
      "learning_rate": 5.1642856696305575e-05,
      "loss": 0.2467,
      "step": 1980
    },
    {
      "epoch": 1.9815573129942796,
      "grad_norm": 0.2515805959701538,
      "learning_rate": 5.155112981266422e-05,
      "loss": 0.2814,
      "step": 1981
    },
    {
      "epoch": 1.9825575943234035,
      "grad_norm": 0.2824844419956207,
      "learning_rate": 5.145945616556921e-05,
      "loss": 0.2689,
      "step": 1982
    },
    {
      "epoch": 1.9835578756525272,
      "grad_norm": 0.3386363089084625,
      "learning_rate": 5.136783585575336e-05,
      "loss": 0.2885,
      "step": 1983
    },
    {
      "epoch": 1.984558156981651,
      "grad_norm": 0.26760348677635193,
      "learning_rate": 5.127626898389075e-05,
      "loss": 0.2797,
      "step": 1984
    },
    {
      "epoch": 1.985558438310775,
      "grad_norm": 0.398053914308548,
      "learning_rate": 5.118475565059691e-05,
      "loss": 0.3142,
      "step": 1985
    },
    {
      "epoch": 1.9865587196398988,
      "grad_norm": 0.29019638895988464,
      "learning_rate": 5.109329595642829e-05,
      "loss": 0.2706,
      "step": 1986
    },
    {
      "epoch": 1.9875590009690225,
      "grad_norm": 0.2805420160293579,
      "learning_rate": 5.1001890001882734e-05,
      "loss": 0.2422,
      "step": 1987
    },
    {
      "epoch": 1.9885592822981464,
      "grad_norm": 0.25784286856651306,
      "learning_rate": 5.091053788739878e-05,
      "loss": 0.2504,
      "step": 1988
    },
    {
      "epoch": 1.98955956362727,
      "grad_norm": 0.3804432451725006,
      "learning_rate": 5.081923971335582e-05,
      "loss": 0.2698,
      "step": 1989
    },
    {
      "epoch": 1.990559844956394,
      "grad_norm": 0.27753493189811707,
      "learning_rate": 5.072799558007415e-05,
      "loss": 0.291,
      "step": 1990
    },
    {
      "epoch": 1.9915601262855178,
      "grad_norm": 0.3878805339336395,
      "learning_rate": 5.063680558781445e-05,
      "loss": 0.298,
      "step": 1991
    },
    {
      "epoch": 1.9925604076146417,
      "grad_norm": 0.2693289816379547,
      "learning_rate": 5.0545669836778144e-05,
      "loss": 0.2591,
      "step": 1992
    },
    {
      "epoch": 1.9935606889437656,
      "grad_norm": 0.30663907527923584,
      "learning_rate": 5.045458842710684e-05,
      "loss": 0.2911,
      "step": 1993
    },
    {
      "epoch": 1.9945609702728893,
      "grad_norm": 0.39006543159484863,
      "learning_rate": 5.036356145888263e-05,
      "loss": 0.2844,
      "step": 1994
    },
    {
      "epoch": 1.995561251602013,
      "grad_norm": 0.2629064917564392,
      "learning_rate": 5.0272589032127594e-05,
      "loss": 0.2768,
      "step": 1995
    },
    {
      "epoch": 1.9965615329311368,
      "grad_norm": 0.2503482699394226,
      "learning_rate": 5.0181671246804064e-05,
      "loss": 0.234,
      "step": 1996
    },
    {
      "epoch": 1.9975618142602607,
      "grad_norm": 0.3425801396369934,
      "learning_rate": 5.009080820281415e-05,
      "loss": 0.2608,
      "step": 1997
    },
    {
      "epoch": 1.9985620955893846,
      "grad_norm": 0.3080214858055115,
      "learning_rate": 5.000000000000002e-05,
      "loss": 0.2854,
      "step": 1998
    },
    {
      "epoch": 1.9995623769185085,
      "grad_norm": 0.2668212652206421,
      "learning_rate": 4.990924673814336e-05,
      "loss": 0.2681,
      "step": 1999
    },
    {
      "epoch": 2.0005626582476324,
      "grad_norm": 0.27917543053627014,
      "learning_rate": 4.981854851696568e-05,
      "loss": 0.251,
      "step": 2000
    }
  ],
  "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": 2.1426628789960704e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}