{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9887005649717513,
  "eval_steps": 500,
  "global_step": 154,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0129136400322841,
      "grad_norm": 26.062711090700205,
      "learning_rate": 1.25e-07,
      "loss": 0.3676,
      "step": 1
    },
    {
      "epoch": 0.0258272800645682,
      "grad_norm": 25.492701005475226,
      "learning_rate": 2.5e-07,
      "loss": 0.3636,
      "step": 2
    },
    {
      "epoch": 0.0387409200968523,
      "grad_norm": 28.061770353754802,
      "learning_rate": 3.75e-07,
      "loss": 0.3788,
      "step": 3
    },
    {
      "epoch": 0.0516545601291364,
      "grad_norm": 28.99654186845467,
      "learning_rate": 5e-07,
      "loss": 0.395,
      "step": 4
    },
    {
      "epoch": 0.0645682001614205,
      "grad_norm": 20.935596355684492,
      "learning_rate": 6.249999999999999e-07,
      "loss": 0.3741,
      "step": 5
    },
    {
      "epoch": 0.0774818401937046,
      "grad_norm": 26.035750645230728,
      "learning_rate": 7.5e-07,
      "loss": 0.3701,
      "step": 6
    },
    {
      "epoch": 0.0903954802259887,
      "grad_norm": 36.159284168135784,
      "learning_rate": 8.75e-07,
      "loss": 0.3642,
      "step": 7
    },
    {
      "epoch": 0.1033091202582728,
      "grad_norm": 32.251355945752145,
      "learning_rate": 1e-06,
      "loss": 0.3473,
      "step": 8
    },
    {
      "epoch": 0.1162227602905569,
      "grad_norm": 35.98055219357963,
      "learning_rate": 1.125e-06,
      "loss": 0.3728,
      "step": 9
    },
    {
      "epoch": 0.129136400322841,
      "grad_norm": 40.267311648036056,
      "learning_rate": 1.2499999999999999e-06,
      "loss": 0.3472,
      "step": 10
    },
    {
      "epoch": 0.1420500403551251,
      "grad_norm": 54.431376922416256,
      "learning_rate": 1.375e-06,
      "loss": 0.3941,
      "step": 11
    },
    {
      "epoch": 0.1549636803874092,
      "grad_norm": 52.787609103487405,
      "learning_rate": 1.5e-06,
      "loss": 0.3694,
      "step": 12
    },
    {
      "epoch": 0.1678773204196933,
      "grad_norm": 51.470919986790086,
      "learning_rate": 1.625e-06,
      "loss": 0.3587,
      "step": 13
    },
    {
      "epoch": 0.1807909604519774,
      "grad_norm": 44.49429708878717,
      "learning_rate": 1.75e-06,
      "loss": 0.3369,
      "step": 14
    },
    {
      "epoch": 0.1937046004842615,
      "grad_norm": 48.61799720707049,
      "learning_rate": 1.8749999999999998e-06,
      "loss": 0.3811,
      "step": 15
    },
    {
      "epoch": 0.2066182405165456,
      "grad_norm": 41.649896181802205,
      "learning_rate": 2e-06,
      "loss": 0.3911,
      "step": 16
    },
    {
      "epoch": 0.2195318805488297,
      "grad_norm": 42.44898868701815,
      "learning_rate": 1.999740884841349e-06,
      "loss": 0.387,
      "step": 17
    },
    {
      "epoch": 0.2324455205811138,
      "grad_norm": 41.183443431747065,
      "learning_rate": 1.9989636736467275e-06,
      "loss": 0.3946,
      "step": 18
    },
    {
      "epoch": 0.2453591606133979,
      "grad_norm": 34.50686543223266,
      "learning_rate": 1.9976687691905393e-06,
      "loss": 0.4088,
      "step": 19
    },
    {
      "epoch": 0.258272800645682,
      "grad_norm": 27.279515728605386,
      "learning_rate": 1.995856842531531e-06,
      "loss": 0.3724,
      "step": 20
    },
    {
      "epoch": 0.2711864406779661,
      "grad_norm": 30.83455319215996,
      "learning_rate": 1.993528832665031e-06,
      "loss": 0.3866,
      "step": 21
    },
    {
      "epoch": 0.2841000807102502,
      "grad_norm": 26.247716082750618,
      "learning_rate": 1.9906859460363304e-06,
      "loss": 0.3521,
      "step": 22
    },
    {
      "epoch": 0.2970137207425343,
      "grad_norm": 27.831795874032476,
      "learning_rate": 1.9873296559154695e-06,
      "loss": 0.4159,
      "step": 23
    },
    {
      "epoch": 0.3099273607748184,
      "grad_norm": 28.5584628971039,
      "learning_rate": 1.983461701633742e-06,
      "loss": 0.3767,
      "step": 24
    },
    {
      "epoch": 0.3228410008071025,
      "grad_norm": 21.378018038217974,
      "learning_rate": 1.979084087682323e-06,
      "loss": 0.3304,
      "step": 25
    },
    {
      "epoch": 0.3357546408393866,
      "grad_norm": 23.044867887515178,
      "learning_rate": 1.9741990826734792e-06,
      "loss": 0.3705,
      "step": 26
    },
    {
      "epoch": 0.3486682808716707,
      "grad_norm": 26.947778108928976,
      "learning_rate": 1.968809218164906e-06,
      "loss": 0.404,
      "step": 27
    },
    {
      "epoch": 0.3615819209039548,
      "grad_norm": 22.039479721325353,
      "learning_rate": 1.9629172873477994e-06,
      "loss": 0.3365,
      "step": 28
    },
    {
      "epoch": 0.3744955609362389,
      "grad_norm": 21.274934357850633,
      "learning_rate": 1.956526343599335e-06,
      "loss": 0.3306,
      "step": 29
    },
    {
      "epoch": 0.387409200968523,
      "grad_norm": 23.40835330420104,
      "learning_rate": 1.949639698900319e-06,
      "loss": 0.3246,
      "step": 30
    },
    {
      "epoch": 0.4003228410008071,
      "grad_norm": 26.03684733839572,
      "learning_rate": 1.9422609221188204e-06,
      "loss": 0.3121,
      "step": 31
    },
    {
      "epoch": 0.4132364810330912,
      "grad_norm": 22.817733184810667,
      "learning_rate": 1.9343938371606708e-06,
      "loss": 0.363,
      "step": 32
    },
    {
      "epoch": 0.4261501210653753,
      "grad_norm": 27.888538264779257,
      "learning_rate": 1.926042520987805e-06,
      "loss": 0.3599,
      "step": 33
    },
    {
      "epoch": 0.4390637610976594,
      "grad_norm": 23.664544295074048,
      "learning_rate": 1.9172113015054528e-06,
      "loss": 0.3259,
      "step": 34
    },
    {
      "epoch": 0.4519774011299435,
      "grad_norm": 26.814661596965937,
      "learning_rate": 1.907904755319289e-06,
      "loss": 0.3622,
      "step": 35
    },
    {
      "epoch": 0.4648910411622276,
      "grad_norm": 23.047494151516773,
      "learning_rate": 1.898127705363696e-06,
      "loss": 0.3595,
      "step": 36
    },
    {
      "epoch": 0.4778046811945117,
      "grad_norm": 23.769391103002327,
      "learning_rate": 1.887885218402375e-06,
      "loss": 0.3418,
      "step": 37
    },
    {
      "epoch": 0.4907183212267958,
      "grad_norm": 21.448479890321668,
      "learning_rate": 1.8771826024025943e-06,
      "loss": 0.3112,
      "step": 38
    },
    {
      "epoch": 0.5036319612590799,
      "grad_norm": 21.618299294862208,
      "learning_rate": 1.8660254037844386e-06,
      "loss": 0.2935,
      "step": 39
    },
    {
      "epoch": 0.516545601291364,
      "grad_norm": 23.83881914235937,
      "learning_rate": 1.8544194045464886e-06,
      "loss": 0.3487,
      "step": 40
    },
    {
      "epoch": 0.5294592413236481,
      "grad_norm": 22.34191137757626,
      "learning_rate": 1.8423706192694115e-06,
      "loss": 0.3337,
      "step": 41
    },
    {
      "epoch": 0.5423728813559322,
      "grad_norm": 22.680526606663665,
      "learning_rate": 1.8298852919990251e-06,
      "loss": 0.3326,
      "step": 42
    },
    {
      "epoch": 0.5552865213882163,
      "grad_norm": 21.047157564860804,
      "learning_rate": 1.8169698930104419e-06,
      "loss": 0.3262,
      "step": 43
    },
    {
      "epoch": 0.5682001614205004,
      "grad_norm": 23.210220151582366,
      "learning_rate": 1.8036311154549781e-06,
      "loss": 0.3531,
      "step": 44
    },
    {
      "epoch": 0.5811138014527845,
      "grad_norm": 23.059400786558072,
      "learning_rate": 1.7898758718915585e-06,
      "loss": 0.3042,
      "step": 45
    },
    {
      "epoch": 0.5940274414850686,
      "grad_norm": 20.985554392897924,
      "learning_rate": 1.7757112907044198e-06,
      "loss": 0.3211,
      "step": 46
    },
    {
      "epoch": 0.6069410815173527,
      "grad_norm": 24.381494037046473,
      "learning_rate": 1.7611447124089646e-06,
      "loss": 0.3501,
      "step": 47
    },
    {
      "epoch": 0.6198547215496368,
      "grad_norm": 21.163889567259087,
      "learning_rate": 1.7461836858476855e-06,
      "loss": 0.3407,
      "step": 48
    },
    {
      "epoch": 0.632768361581921,
      "grad_norm": 17.673857406454214,
      "learning_rate": 1.7308359642781241e-06,
      "loss": 0.2912,
      "step": 49
    },
    {
      "epoch": 0.645682001614205,
      "grad_norm": 20.851302876331236,
      "learning_rate": 1.7151095013548993e-06,
      "loss": 0.311,
      "step": 50
    },
    {
      "epoch": 0.6585956416464891,
      "grad_norm": 20.07386513124459,
      "learning_rate": 1.699012447007882e-06,
      "loss": 0.2891,
      "step": 51
    },
    {
      "epoch": 0.6715092816787732,
      "grad_norm": 16.654008916412547,
      "learning_rate": 1.682553143218654e-06,
      "loss": 0.2724,
      "step": 52
    },
    {
      "epoch": 0.6844229217110573,
      "grad_norm": 20.770464397707823,
      "learning_rate": 1.6657401196974403e-06,
      "loss": 0.332,
      "step": 53
    },
    {
      "epoch": 0.6973365617433414,
      "grad_norm": 22.094352414249297,
      "learning_rate": 1.648582089462756e-06,
      "loss": 0.344,
      "step": 54
    },
    {
      "epoch": 0.7102502017756255,
      "grad_norm": 20.494531642603356,
      "learning_rate": 1.6310879443260529e-06,
      "loss": 0.3041,
      "step": 55
    },
    {
      "epoch": 0.7231638418079096,
      "grad_norm": 19.8177079163147,
      "learning_rate": 1.6132667502837162e-06,
      "loss": 0.3162,
      "step": 56
    },
    {
      "epoch": 0.7360774818401937,
      "grad_norm": 19.049318629916456,
      "learning_rate": 1.5951277428187897e-06,
      "loss": 0.3094,
      "step": 57
    },
    {
      "epoch": 0.7489911218724778,
      "grad_norm": 21.597536610054288,
      "learning_rate": 1.5766803221148673e-06,
      "loss": 0.3325,
      "step": 58
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 20.70892358284607,
      "learning_rate": 1.5579340481846335e-06,
      "loss": 0.3397,
      "step": 59
    },
    {
      "epoch": 0.774818401937046,
      "grad_norm": 20.308915238405316,
      "learning_rate": 1.5388986359155755e-06,
      "loss": 0.3265,
      "step": 60
    },
    {
      "epoch": 0.7877320419693301,
      "grad_norm": 22.627261985685145,
      "learning_rate": 1.5195839500354335e-06,
      "loss": 0.329,
      "step": 61
    },
    {
      "epoch": 0.8006456820016142,
      "grad_norm": 19.565397419582293,
      "learning_rate": 1.5e-06,
      "loss": 0.3167,
      "step": 62
    },
    {
      "epoch": 0.8135593220338984,
      "grad_norm": 18.907828874624407,
      "learning_rate": 1.4801569348059155e-06,
      "loss": 0.3146,
      "step": 63
    },
    {
      "epoch": 0.8264729620661824,
      "grad_norm": 22.644125199961817,
      "learning_rate": 1.460065037731152e-06,
      "loss": 0.3072,
      "step": 64
    },
    {
      "epoch": 0.8393866020984665,
      "grad_norm": 20.786645375646636,
      "learning_rate": 1.4397347210059057e-06,
      "loss": 0.3159,
      "step": 65
    },
    {
      "epoch": 0.8523002421307506,
      "grad_norm": 18.57141588916772,
      "learning_rate": 1.419176520416664e-06,
      "loss": 0.3103,
      "step": 66
    },
    {
      "epoch": 0.8652138821630347,
      "grad_norm": 17.225037473257153,
      "learning_rate": 1.3984010898462415e-06,
      "loss": 0.3027,
      "step": 67
    },
    {
      "epoch": 0.8781275221953188,
      "grad_norm": 18.21584058115977,
      "learning_rate": 1.3774191957526142e-06,
      "loss": 0.2895,
      "step": 68
    },
    {
      "epoch": 0.8910411622276029,
      "grad_norm": 20.509880420628917,
      "learning_rate": 1.3562417115894168e-06,
      "loss": 0.3098,
      "step": 69
    },
    {
      "epoch": 0.903954802259887,
      "grad_norm": 17.29318336230985,
      "learning_rate": 1.334879612170986e-06,
      "loss": 0.2954,
      "step": 70
    },
    {
      "epoch": 0.9168684422921711,
      "grad_norm": 18.377551002595997,
      "learning_rate": 1.3133439679848822e-06,
      "loss": 0.301,
      "step": 71
    },
    {
      "epoch": 0.9297820823244553,
      "grad_norm": 26.11617966128903,
      "learning_rate": 1.2916459394548249e-06,
      "loss": 0.3493,
      "step": 72
    },
    {
      "epoch": 0.9426957223567393,
      "grad_norm": 19.11873065402881,
      "learning_rate": 1.2697967711570242e-06,
      "loss": 0.3055,
      "step": 73
    },
    {
      "epoch": 0.9556093623890234,
      "grad_norm": 20.15187293059899,
      "learning_rate": 1.2478077859929e-06,
      "loss": 0.3267,
      "step": 74
    },
    {
      "epoch": 0.9685230024213075,
      "grad_norm": 17.356504787996283,
      "learning_rate": 1.2256903793212105e-06,
      "loss": 0.3,
      "step": 75
    },
    {
      "epoch": 0.9814366424535916,
      "grad_norm": 20.116822076601576,
      "learning_rate": 1.203456013052634e-06,
      "loss": 0.3139,
      "step": 76
    },
    {
      "epoch": 0.9943502824858758,
      "grad_norm": 20.23077529953701,
      "learning_rate": 1.1811162097098558e-06,
      "loss": 0.2941,
      "step": 77
    },
    {
      "epoch": 1.0072639225181599,
      "grad_norm": 14.472243001169707,
      "learning_rate": 1.1586825464562514e-06,
      "loss": 0.217,
      "step": 78
    },
    {
      "epoch": 1.020177562550444,
      "grad_norm": 11.1760098722801,
      "learning_rate": 1.1361666490962467e-06,
      "loss": 0.1391,
      "step": 79
    },
    {
      "epoch": 1.033091202582728,
      "grad_norm": 12.929875937980091,
      "learning_rate": 1.1135801860504748e-06,
      "loss": 0.144,
      "step": 80
    },
    {
      "epoch": 1.0460048426150121,
      "grad_norm": 14.401479453367932,
      "learning_rate": 1.0909348623088471e-06,
      "loss": 0.1525,
      "step": 81
    },
    {
      "epoch": 1.0589184826472962,
      "grad_norm": 11.466941187008539,
      "learning_rate": 1.068242413364671e-06,
      "loss": 0.1492,
      "step": 82
    },
    {
      "epoch": 1.0718321226795804,
      "grad_norm": 13.844595316123245,
      "learning_rate": 1.0455145991329637e-06,
      "loss": 0.1317,
      "step": 83
    },
    {
      "epoch": 1.0847457627118644,
      "grad_norm": 14.390882441536627,
      "learning_rate": 1.0227631978561055e-06,
      "loss": 0.1574,
      "step": 84
    },
    {
      "epoch": 1.0976594027441484,
      "grad_norm": 13.517598225715695,
      "learning_rate": 1e-06,
      "loss": 0.1409,
      "step": 85
    },
    {
      "epoch": 1.1105730427764327,
      "grad_norm": 13.993115366941224,
      "learning_rate": 9.772368021438942e-07,
      "loss": 0.1464,
      "step": 86
    },
    {
      "epoch": 1.1234866828087167,
      "grad_norm": 12.2423524228492,
      "learning_rate": 9.544854008670366e-07,
      "loss": 0.1207,
      "step": 87
    },
    {
      "epoch": 1.136400322841001,
      "grad_norm": 14.122285998417135,
      "learning_rate": 9.317575866353291e-07,
      "loss": 0.1251,
      "step": 88
    },
    {
      "epoch": 1.149313962873285,
      "grad_norm": 15.560999029473845,
      "learning_rate": 9.09065137691153e-07,
      "loss": 0.1209,
      "step": 89
    },
    {
      "epoch": 1.162227602905569,
      "grad_norm": 13.507106399175507,
      "learning_rate": 8.86419813949525e-07,
      "loss": 0.1091,
      "step": 90
    },
    {
      "epoch": 1.1751412429378532,
      "grad_norm": 13.891730510521498,
      "learning_rate": 8.638333509037535e-07,
      "loss": 0.1256,
      "step": 91
    },
    {
      "epoch": 1.1880548829701372,
      "grad_norm": 14.992783015617869,
      "learning_rate": 8.413174535437487e-07,
      "loss": 0.137,
      "step": 92
    },
    {
      "epoch": 1.2009685230024214,
      "grad_norm": 15.910575644376518,
      "learning_rate": 8.188837902901441e-07,
      "loss": 0.1347,
      "step": 93
    },
    {
      "epoch": 1.2138821630347054,
      "grad_norm": 19.326658299865034,
      "learning_rate": 7.965439869473663e-07,
      "loss": 0.1473,
      "step": 94
    },
    {
      "epoch": 1.2267958030669894,
      "grad_norm": 21.139159288452323,
      "learning_rate": 7.743096206787893e-07,
      "loss": 0.1449,
      "step": 95
    },
    {
      "epoch": 1.2397094430992737,
      "grad_norm": 19.57794361709339,
      "learning_rate": 7.521922140071003e-07,
      "loss": 0.1429,
      "step": 96
    },
    {
      "epoch": 1.2526230831315577,
      "grad_norm": 18.770894293602584,
      "learning_rate": 7.302032288429756e-07,
      "loss": 0.1533,
      "step": 97
    },
    {
      "epoch": 1.2655367231638417,
      "grad_norm": 19.148357746011182,
      "learning_rate": 7.083540605451749e-07,
      "loss": 0.164,
      "step": 98
    },
    {
      "epoch": 1.278450363196126,
      "grad_norm": 15.077187564406879,
      "learning_rate": 6.866560320151178e-07,
      "loss": 0.1404,
      "step": 99
    },
    {
      "epoch": 1.29136400322841,
      "grad_norm": 16.058568106787366,
      "learning_rate": 6.651203878290138e-07,
      "loss": 0.1327,
      "step": 100
    },
    {
      "epoch": 1.3042776432606942,
      "grad_norm": 14.029168964991912,
      "learning_rate": 6.437582884105834e-07,
      "loss": 0.136,
      "step": 101
    },
    {
      "epoch": 1.3171912832929782,
      "grad_norm": 15.10100507251639,
      "learning_rate": 6.225808042473857e-07,
      "loss": 0.1149,
      "step": 102
    },
    {
      "epoch": 1.3301049233252624,
      "grad_norm": 11.718976894577146,
      "learning_rate": 6.015989101537586e-07,
      "loss": 0.1207,
      "step": 103
    },
    {
      "epoch": 1.3430185633575464,
      "grad_norm": 12.627764504381856,
      "learning_rate": 5.808234795833362e-07,
      "loss": 0.1081,
      "step": 104
    },
    {
      "epoch": 1.3559322033898304,
      "grad_norm": 15.041188760891716,
      "learning_rate": 5.602652789940941e-07,
      "loss": 0.1262,
      "step": 105
    },
    {
      "epoch": 1.3688458434221147,
      "grad_norm": 21.09158607094679,
      "learning_rate": 5.399349622688478e-07,
      "loss": 0.1314,
      "step": 106
    },
    {
      "epoch": 1.3817594834543987,
      "grad_norm": 20.9326453419062,
      "learning_rate": 5.198430651940845e-07,
      "loss": 0.1346,
      "step": 107
    },
    {
      "epoch": 1.3946731234866827,
      "grad_norm": 12.206782639953575,
      "learning_rate": 5.000000000000002e-07,
      "loss": 0.1157,
      "step": 108
    },
    {
      "epoch": 1.407586763518967,
      "grad_norm": 13.75893208112109,
      "learning_rate": 4.804160499645667e-07,
      "loss": 0.1384,
      "step": 109
    },
    {
      "epoch": 1.420500403551251,
      "grad_norm": 13.597298102282235,
      "learning_rate": 4.6110136408442445e-07,
      "loss": 0.1137,
      "step": 110
    },
    {
      "epoch": 1.4334140435835352,
      "grad_norm": 13.538763536743675,
      "learning_rate": 4.4206595181536664e-07,
      "loss": 0.1203,
      "step": 111
    },
    {
      "epoch": 1.4463276836158192,
      "grad_norm": 12.774053600703025,
      "learning_rate": 4.233196778851329e-07,
      "loss": 0.1122,
      "step": 112
    },
    {
      "epoch": 1.4592413236481034,
      "grad_norm": 12.870680344192039,
      "learning_rate": 4.0487225718121045e-07,
      "loss": 0.1228,
      "step": 113
    },
    {
      "epoch": 1.4721549636803875,
      "grad_norm": 15.02538486601625,
      "learning_rate": 3.867332497162835e-07,
      "loss": 0.1615,
      "step": 114
    },
    {
      "epoch": 1.4850686037126715,
      "grad_norm": 12.30281015942022,
      "learning_rate": 3.6891205567394746e-07,
      "loss": 0.099,
      "step": 115
    },
    {
      "epoch": 1.4979822437449557,
      "grad_norm": 10.419235242267765,
      "learning_rate": 3.5141791053724404e-07,
      "loss": 0.1096,
      "step": 116
    },
    {
      "epoch": 1.5108958837772397,
      "grad_norm": 11.16767240028713,
      "learning_rate": 3.3425988030255945e-07,
      "loss": 0.1199,
      "step": 117
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 13.158987202921047,
      "learning_rate": 3.174468567813461e-07,
      "loss": 0.1324,
      "step": 118
    },
    {
      "epoch": 1.536723163841808,
      "grad_norm": 11.844158983050455,
      "learning_rate": 3.009875529921181e-07,
      "loss": 0.1106,
      "step": 119
    },
    {
      "epoch": 1.549636803874092,
      "grad_norm": 14.439077228213756,
      "learning_rate": 2.848904986451005e-07,
      "loss": 0.117,
      "step": 120
    },
    {
      "epoch": 1.562550443906376,
      "grad_norm": 12.407600386041583,
      "learning_rate": 2.6916403572187587e-07,
      "loss": 0.1159,
      "step": 121
    },
    {
      "epoch": 1.5754640839386602,
      "grad_norm": 13.151590889043437,
      "learning_rate": 2.538163141523145e-07,
      "loss": 0.1236,
      "step": 122
    },
    {
      "epoch": 1.5883777239709445,
      "grad_norm": 12.342415661391085,
      "learning_rate": 2.3885528759103536e-07,
      "loss": 0.1191,
      "step": 123
    },
    {
      "epoch": 1.6012913640032282,
      "grad_norm": 25.46717666484534,
      "learning_rate": 2.2428870929558007e-07,
      "loss": 0.1464,
      "step": 124
    },
    {
      "epoch": 1.6142050040355125,
      "grad_norm": 13.835165114851305,
      "learning_rate": 2.101241281084416e-07,
      "loss": 0.114,
      "step": 125
    },
    {
      "epoch": 1.6271186440677967,
      "grad_norm": 12.278805533417628,
      "learning_rate": 1.9636888454502177e-07,
      "loss": 0.1095,
      "step": 126
    },
    {
      "epoch": 1.6400322841000807,
      "grad_norm": 9.656229159282553,
      "learning_rate": 1.83030106989558e-07,
      "loss": 0.1048,
      "step": 127
    },
    {
      "epoch": 1.6529459241323647,
      "grad_norm": 10.723649324385244,
      "learning_rate": 1.7011470800097495e-07,
      "loss": 0.1144,
      "step": 128
    },
    {
      "epoch": 1.665859564164649,
      "grad_norm": 11.18317780807273,
      "learning_rate": 1.576293807305885e-07,
      "loss": 0.1124,
      "step": 129
    },
    {
      "epoch": 1.678773204196933,
      "grad_norm": 19.22525610881903,
      "learning_rate": 1.4558059545351142e-07,
      "loss": 0.1185,
      "step": 130
    },
    {
      "epoch": 1.691686844229217,
      "grad_norm": 13.840544386244854,
      "learning_rate": 1.3397459621556128e-07,
      "loss": 0.1172,
      "step": 131
    },
    {
      "epoch": 1.7046004842615012,
      "grad_norm": 9.863106547796383,
      "learning_rate": 1.2281739759740574e-07,
      "loss": 0.1123,
      "step": 132
    },
    {
      "epoch": 1.7175141242937855,
      "grad_norm": 12.249146236722565,
      "learning_rate": 1.1211478159762478e-07,
      "loss": 0.1026,
      "step": 133
    },
    {
      "epoch": 1.7304277643260693,
      "grad_norm": 19.53823199805843,
      "learning_rate": 1.0187229463630398e-07,
      "loss": 0.1245,
      "step": 134
    },
    {
      "epoch": 1.7433414043583535,
      "grad_norm": 13.813268314685585,
      "learning_rate": 9.209524468071095e-08,
      "loss": 0.1259,
      "step": 135
    },
    {
      "epoch": 1.7562550443906377,
      "grad_norm": 13.002603579106614,
      "learning_rate": 8.278869849454717e-08,
      "loss": 0.126,
      "step": 136
    },
    {
      "epoch": 1.7691686844229217,
      "grad_norm": 11.790731140563368,
      "learning_rate": 7.395747901219473e-08,
      "loss": 0.1163,
      "step": 137
    },
    {
      "epoch": 1.7820823244552058,
      "grad_norm": 15.611451697022517,
      "learning_rate": 6.560616283932896e-08,
      "loss": 0.1226,
      "step": 138
    },
    {
      "epoch": 1.79499596448749,
      "grad_norm": 24.410099015402505,
      "learning_rate": 5.77390778811796e-08,
      "loss": 0.1155,
      "step": 139
    },
    {
      "epoch": 1.807909604519774,
      "grad_norm": 9.829584476780473,
      "learning_rate": 5.036030109968081e-08,
      "loss": 0.0972,
      "step": 140
    },
    {
      "epoch": 1.820823244552058,
      "grad_norm": 14.542859998120877,
      "learning_rate": 4.347365640066525e-08,
      "loss": 0.1126,
      "step": 141
    },
    {
      "epoch": 1.8337368845843423,
      "grad_norm": 20.231191724041285,
      "learning_rate": 3.708271265220087e-08,
      "loss": 0.1175,
      "step": 142
    },
    {
      "epoch": 1.8466505246166263,
      "grad_norm": 12.407632368224395,
      "learning_rate": 3.119078183509372e-08,
      "loss": 0.1149,
      "step": 143
    },
    {
      "epoch": 1.8595641646489103,
      "grad_norm": 14.44661290494769,
      "learning_rate": 2.580091732652101e-08,
      "loss": 0.1181,
      "step": 144
    },
    {
      "epoch": 1.8724778046811945,
      "grad_norm": 18.79024412476877,
      "learning_rate": 2.0915912317677088e-08,
      "loss": 0.1475,
      "step": 145
    },
    {
      "epoch": 1.8853914447134787,
      "grad_norm": 15.376557680187412,
      "learning_rate": 1.6538298366257974e-08,
      "loss": 0.1255,
      "step": 146
    },
    {
      "epoch": 1.8983050847457628,
      "grad_norm": 15.33944835023639,
      "learning_rate": 1.2670344084530382e-08,
      "loss": 0.1034,
      "step": 147
    },
    {
      "epoch": 1.9112187247780468,
      "grad_norm": 13.112291504434255,
      "learning_rate": 9.314053963669244e-09,
      "loss": 0.1086,
      "step": 148
    },
    {
      "epoch": 1.924132364810331,
      "grad_norm": 12.045688670027003,
      "learning_rate": 6.471167334968886e-09,
      "loss": 0.1096,
      "step": 149
    },
    {
      "epoch": 1.937046004842615,
      "grad_norm": 15.127302877916637,
      "learning_rate": 4.143157468468716e-09,
      "loss": 0.1126,
      "step": 150
    },
    {
      "epoch": 1.949959644874899,
      "grad_norm": 13.774563816455347,
      "learning_rate": 2.331230809460738e-09,
      "loss": 0.1245,
      "step": 151
    },
    {
      "epoch": 1.9628732849071833,
      "grad_norm": 32.82358453731887,
      "learning_rate": 1.036326353272443e-09,
      "loss": 0.1181,
      "step": 152
    },
    {
      "epoch": 1.9757869249394673,
      "grad_norm": 14.92150992963698,
      "learning_rate": 2.5911515865084663e-10,
      "loss": 0.1143,
      "step": 153
    },
    {
      "epoch": 1.9887005649717513,
      "grad_norm": 13.25981887405618,
      "learning_rate": 0.0,
      "loss": 0.1276,
      "step": 154
    },
    {
      "epoch": 1.9887005649717513,
      "step": 154,
      "total_flos": 72055202119680.0,
      "train_loss": 0.2331618887747263,
      "train_runtime": 3106.2852,
      "train_samples_per_second": 12.759,
      "train_steps_per_second": 0.05
    }
  ],
  "logging_steps": 1,
  "max_steps": 154,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 72055202119680.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}