{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 500,
  "global_step": 11140,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.18,
      "grad_norm": 6.054772853851318,
      "learning_rate": 1.9820466786355476e-05,
      "loss": 0.5721,
      "step": 100
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.810357093811035,
      "learning_rate": 1.9640933572710953e-05,
      "loss": 0.3846,
      "step": 200
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.026122093200684,
      "learning_rate": 1.9461400359066428e-05,
      "loss": 0.3019,
      "step": 300
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.95788288116455,
      "learning_rate": 1.9281867145421905e-05,
      "loss": 0.3071,
      "step": 400
    },
    {
      "epoch": 0.9,
      "grad_norm": 15.329608917236328,
      "learning_rate": 1.910233393177738e-05,
      "loss": 0.2907,
      "step": 500
    },
    {
      "epoch": 1.08,
      "grad_norm": 12.364314079284668,
      "learning_rate": 1.8922800718132857e-05,
      "loss": 0.2646,
      "step": 600
    },
    {
      "epoch": 1.26,
      "grad_norm": 14.555986404418945,
      "learning_rate": 1.874326750448833e-05,
      "loss": 0.2097,
      "step": 700
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.199421405792236,
      "learning_rate": 1.8563734290843805e-05,
      "loss": 0.2389,
      "step": 800
    },
    {
      "epoch": 1.62,
      "grad_norm": 22.485984802246094,
      "learning_rate": 1.8384201077199283e-05,
      "loss": 0.2019,
      "step": 900
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.688520431518555,
      "learning_rate": 1.820466786355476e-05,
      "loss": 0.2594,
      "step": 1000
    },
    {
      "epoch": 1.97,
      "grad_norm": 18.052719116210938,
      "learning_rate": 1.8025134649910235e-05,
      "loss": 0.2044,
      "step": 1100
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.7371789216995239,
      "learning_rate": 1.7845601436265712e-05,
      "loss": 0.1551,
      "step": 1200
    },
    {
      "epoch": 2.33,
      "grad_norm": 20.938648223876953,
      "learning_rate": 1.7666068222621186e-05,
      "loss": 0.1463,
      "step": 1300
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.25227147340774536,
      "learning_rate": 1.748653500897666e-05,
      "loss": 0.1493,
      "step": 1400
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.27634137868881226,
      "learning_rate": 1.7307001795332138e-05,
      "loss": 0.1649,
      "step": 1500
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.2588340938091278,
      "learning_rate": 1.7127468581687616e-05,
      "loss": 0.1521,
      "step": 1600
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.05350634083151817,
      "learning_rate": 1.694793536804309e-05,
      "loss": 0.1343,
      "step": 1700
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.02972230687737465,
      "learning_rate": 1.6768402154398564e-05,
      "loss": 0.1068,
      "step": 1800
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.09572970867156982,
      "learning_rate": 1.658886894075404e-05,
      "loss": 0.1151,
      "step": 1900
    },
    {
      "epoch": 3.59,
      "grad_norm": 21.431325912475586,
      "learning_rate": 1.6409335727109516e-05,
      "loss": 0.1073,
      "step": 2000
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.4688669443130493,
      "learning_rate": 1.6229802513464993e-05,
      "loss": 0.1098,
      "step": 2100
    },
    {
      "epoch": 3.95,
      "grad_norm": 19.461355209350586,
      "learning_rate": 1.6050269299820467e-05,
      "loss": 0.1238,
      "step": 2200
    },
    {
      "epoch": 4.13,
      "grad_norm": 6.33543586730957,
      "learning_rate": 1.5870736086175945e-05,
      "loss": 0.0934,
      "step": 2300
    },
    {
      "epoch": 4.31,
      "grad_norm": 10.25698184967041,
      "learning_rate": 1.569120287253142e-05,
      "loss": 0.068,
      "step": 2400
    },
    {
      "epoch": 4.49,
      "grad_norm": 0.05421575903892517,
      "learning_rate": 1.5511669658886893e-05,
      "loss": 0.0767,
      "step": 2500
    },
    {
      "epoch": 4.67,
      "grad_norm": 0.04917303845286369,
      "learning_rate": 1.533213644524237e-05,
      "loss": 0.1053,
      "step": 2600
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.722901463508606,
      "learning_rate": 1.5152603231597847e-05,
      "loss": 0.0513,
      "step": 2700
    },
    {
      "epoch": 5.03,
      "grad_norm": 28.548158645629883,
      "learning_rate": 1.4973070017953321e-05,
      "loss": 0.0611,
      "step": 2800
    },
    {
      "epoch": 5.21,
      "grad_norm": 1.0562294721603394,
      "learning_rate": 1.4793536804308799e-05,
      "loss": 0.0543,
      "step": 2900
    },
    {
      "epoch": 5.39,
      "grad_norm": 0.011326675303280354,
      "learning_rate": 1.4614003590664274e-05,
      "loss": 0.0491,
      "step": 3000
    },
    {
      "epoch": 5.57,
      "grad_norm": 0.009987740777432919,
      "learning_rate": 1.4434470377019749e-05,
      "loss": 0.0567,
      "step": 3100
    },
    {
      "epoch": 5.75,
      "grad_norm": 26.81354331970215,
      "learning_rate": 1.4254937163375226e-05,
      "loss": 0.0641,
      "step": 3200
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.042816389352083206,
      "learning_rate": 1.4075403949730702e-05,
      "loss": 0.0488,
      "step": 3300
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.014200630597770214,
      "learning_rate": 1.3895870736086176e-05,
      "loss": 0.0374,
      "step": 3400
    },
    {
      "epoch": 6.28,
      "grad_norm": 0.004939161241054535,
      "learning_rate": 1.3716337522441652e-05,
      "loss": 0.0393,
      "step": 3500
    },
    {
      "epoch": 6.46,
      "grad_norm": 39.326534271240234,
      "learning_rate": 1.353680430879713e-05,
      "loss": 0.0404,
      "step": 3600
    },
    {
      "epoch": 6.64,
      "grad_norm": 1.0669581890106201,
      "learning_rate": 1.3357271095152604e-05,
      "loss": 0.03,
      "step": 3700
    },
    {
      "epoch": 6.82,
      "grad_norm": 0.11205600947141647,
      "learning_rate": 1.317773788150808e-05,
      "loss": 0.0416,
      "step": 3800
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.005933618638664484,
      "learning_rate": 1.2998204667863557e-05,
      "loss": 0.0498,
      "step": 3900
    },
    {
      "epoch": 7.18,
      "grad_norm": 0.0025591508019715548,
      "learning_rate": 1.2818671454219031e-05,
      "loss": 0.0214,
      "step": 4000
    },
    {
      "epoch": 7.36,
      "grad_norm": 0.0021508955396711826,
      "learning_rate": 1.2639138240574507e-05,
      "loss": 0.0346,
      "step": 4100
    },
    {
      "epoch": 7.54,
      "grad_norm": 0.002263248898088932,
      "learning_rate": 1.2459605026929983e-05,
      "loss": 0.0315,
      "step": 4200
    },
    {
      "epoch": 7.72,
      "grad_norm": 0.0016491829883307219,
      "learning_rate": 1.2280071813285459e-05,
      "loss": 0.0238,
      "step": 4300
    },
    {
      "epoch": 7.9,
      "grad_norm": 0.0034240155946463346,
      "learning_rate": 1.2100538599640935e-05,
      "loss": 0.0319,
      "step": 4400
    },
    {
      "epoch": 8.08,
      "grad_norm": 0.002759799361228943,
      "learning_rate": 1.1921005385996409e-05,
      "loss": 0.0487,
      "step": 4500
    },
    {
      "epoch": 8.26,
      "grad_norm": 34.639888763427734,
      "learning_rate": 1.1741472172351887e-05,
      "loss": 0.053,
      "step": 4600
    },
    {
      "epoch": 8.44,
      "grad_norm": 0.001209100242704153,
      "learning_rate": 1.1561938958707361e-05,
      "loss": 0.0143,
      "step": 4700
    },
    {
      "epoch": 8.62,
      "grad_norm": 0.0022134396713227034,
      "learning_rate": 1.1382405745062837e-05,
      "loss": 0.0291,
      "step": 4800
    },
    {
      "epoch": 8.8,
      "grad_norm": 0.06826081871986389,
      "learning_rate": 1.1202872531418314e-05,
      "loss": 0.0301,
      "step": 4900
    },
    {
      "epoch": 8.98,
      "grad_norm": 0.0025578713975846767,
      "learning_rate": 1.1023339317773789e-05,
      "loss": 0.0194,
      "step": 5000
    },
    {
      "epoch": 9.16,
      "grad_norm": 0.01461367029696703,
      "learning_rate": 1.0843806104129264e-05,
      "loss": 0.0152,
      "step": 5100
    },
    {
      "epoch": 9.34,
      "grad_norm": 0.0008313595899380744,
      "learning_rate": 1.0664272890484742e-05,
      "loss": 0.0163,
      "step": 5200
    },
    {
      "epoch": 9.52,
      "grad_norm": 1.5908000469207764,
      "learning_rate": 1.0484739676840216e-05,
      "loss": 0.0352,
      "step": 5300
    },
    {
      "epoch": 9.69,
      "grad_norm": 0.0016398399602621794,
      "learning_rate": 1.0305206463195692e-05,
      "loss": 0.0268,
      "step": 5400
    },
    {
      "epoch": 9.87,
      "grad_norm": 0.005051742307841778,
      "learning_rate": 1.0125673249551166e-05,
      "loss": 0.0244,
      "step": 5500
    },
    {
      "epoch": 10.05,
      "grad_norm": 0.11044131964445114,
      "learning_rate": 9.946140035906644e-06,
      "loss": 0.0318,
      "step": 5600
    },
    {
      "epoch": 10.23,
      "grad_norm": 0.02652016654610634,
      "learning_rate": 9.76660682226212e-06,
      "loss": 0.0224,
      "step": 5700
    },
    {
      "epoch": 10.41,
      "grad_norm": 0.0051875789649784565,
      "learning_rate": 9.587073608617595e-06,
      "loss": 0.0086,
      "step": 5800
    },
    {
      "epoch": 10.59,
      "grad_norm": 0.0023671940434724092,
      "learning_rate": 9.40754039497307e-06,
      "loss": 0.0061,
      "step": 5900
    },
    {
      "epoch": 10.77,
      "grad_norm": 0.14054147899150848,
      "learning_rate": 9.228007181328547e-06,
      "loss": 0.0222,
      "step": 6000
    },
    {
      "epoch": 10.95,
      "grad_norm": 0.002135949907824397,
      "learning_rate": 9.048473967684023e-06,
      "loss": 0.0205,
      "step": 6100
    },
    {
      "epoch": 11.13,
      "grad_norm": 0.5219959616661072,
      "learning_rate": 8.868940754039497e-06,
      "loss": 0.0243,
      "step": 6200
    },
    {
      "epoch": 11.31,
      "grad_norm": 0.004496410954743624,
      "learning_rate": 8.689407540394975e-06,
      "loss": 0.0232,
      "step": 6300
    },
    {
      "epoch": 11.49,
      "grad_norm": 0.0012191541027277708,
      "learning_rate": 8.509874326750449e-06,
      "loss": 0.0174,
      "step": 6400
    },
    {
      "epoch": 11.67,
      "grad_norm": 0.0010406944202259183,
      "learning_rate": 8.330341113105925e-06,
      "loss": 0.0002,
      "step": 6500
    },
    {
      "epoch": 11.85,
      "grad_norm": 0.0007079216302372515,
      "learning_rate": 8.1508078994614e-06,
      "loss": 0.0181,
      "step": 6600
    },
    {
      "epoch": 12.03,
      "grad_norm": 0.0031879213638603687,
      "learning_rate": 7.971274685816877e-06,
      "loss": 0.0029,
      "step": 6700
    },
    {
      "epoch": 12.21,
      "grad_norm": 0.07363840192556381,
      "learning_rate": 7.791741472172353e-06,
      "loss": 0.012,
      "step": 6800
    },
    {
      "epoch": 12.39,
      "grad_norm": 0.0074967676773667336,
      "learning_rate": 7.6122082585278276e-06,
      "loss": 0.0297,
      "step": 6900
    },
    {
      "epoch": 12.57,
      "grad_norm": 0.03205731883645058,
      "learning_rate": 7.432675044883304e-06,
      "loss": 0.0172,
      "step": 7000
    },
    {
      "epoch": 12.75,
      "grad_norm": 0.0017446995479986072,
      "learning_rate": 7.25314183123878e-06,
      "loss": 0.0095,
      "step": 7100
    },
    {
      "epoch": 12.93,
      "grad_norm": 62.21705627441406,
      "learning_rate": 7.073608617594255e-06,
      "loss": 0.0216,
      "step": 7200
    },
    {
      "epoch": 13.11,
      "grad_norm": 0.0004949498688802123,
      "learning_rate": 6.894075403949732e-06,
      "loss": 0.0025,
      "step": 7300
    },
    {
      "epoch": 13.29,
      "grad_norm": 1.0483726263046265,
      "learning_rate": 6.714542190305207e-06,
      "loss": 0.0124,
      "step": 7400
    },
    {
      "epoch": 13.46,
      "grad_norm": 0.0011325060622766614,
      "learning_rate": 6.535008976660683e-06,
      "loss": 0.0176,
      "step": 7500
    },
    {
      "epoch": 13.64,
      "grad_norm": 0.0004886816022917628,
      "learning_rate": 6.355475763016159e-06,
      "loss": 0.0005,
      "step": 7600
    },
    {
      "epoch": 13.82,
      "grad_norm": 0.0002447882143314928,
      "learning_rate": 6.175942549371634e-06,
      "loss": 0.0,
      "step": 7700
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.0033621052280068398,
      "learning_rate": 5.99640933572711e-06,
      "loss": 0.0151,
      "step": 7800
    },
    {
      "epoch": 14.18,
      "grad_norm": 0.37218067049980164,
      "learning_rate": 5.8168761220825854e-06,
      "loss": 0.0062,
      "step": 7900
    },
    {
      "epoch": 14.36,
      "grad_norm": 0.00017587828915566206,
      "learning_rate": 5.637342908438061e-06,
      "loss": 0.0004,
      "step": 8000
    },
    {
      "epoch": 14.54,
      "grad_norm": 0.00030748211429454386,
      "learning_rate": 5.457809694793538e-06,
      "loss": 0.0,
      "step": 8100
    },
    {
      "epoch": 14.72,
      "grad_norm": 0.0006639692583121359,
      "learning_rate": 5.278276481149013e-06,
      "loss": 0.0301,
      "step": 8200
    },
    {
      "epoch": 14.9,
      "grad_norm": 0.0006772920023649931,
      "learning_rate": 5.098743267504489e-06,
      "loss": 0.015,
      "step": 8300
    },
    {
      "epoch": 15.08,
      "grad_norm": 0.0017782676732167602,
      "learning_rate": 4.919210053859965e-06,
      "loss": 0.0184,
      "step": 8400
    },
    {
      "epoch": 15.26,
      "grad_norm": 0.0006229325663298368,
      "learning_rate": 4.739676840215441e-06,
      "loss": 0.0053,
      "step": 8500
    },
    {
      "epoch": 15.44,
      "grad_norm": 0.0011186526389792562,
      "learning_rate": 4.560143626570916e-06,
      "loss": 0.0076,
      "step": 8600
    },
    {
      "epoch": 15.62,
      "grad_norm": 0.0009836511453613639,
      "learning_rate": 4.380610412926392e-06,
      "loss": 0.0054,
      "step": 8700
    },
    {
      "epoch": 15.8,
      "grad_norm": 0.002876699436455965,
      "learning_rate": 4.2010771992818675e-06,
      "loss": 0.0011,
      "step": 8800
    },
    {
      "epoch": 15.98,
      "grad_norm": 0.0004506981058511883,
      "learning_rate": 4.021543985637343e-06,
      "loss": 0.0142,
      "step": 8900
    },
    {
      "epoch": 16.16,
      "grad_norm": 143.88809204101562,
      "learning_rate": 3.842010771992819e-06,
      "loss": 0.0136,
      "step": 9000
    },
    {
      "epoch": 16.34,
      "grad_norm": 0.0012065657647326589,
      "learning_rate": 3.6624775583482946e-06,
      "loss": 0.0025,
      "step": 9100
    },
    {
      "epoch": 16.52,
      "grad_norm": 0.0008388046990148723,
      "learning_rate": 3.48294434470377e-06,
      "loss": 0.0054,
      "step": 9200
    },
    {
      "epoch": 16.7,
      "grad_norm": 0.0005097670364193618,
      "learning_rate": 3.3034111310592464e-06,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 16.88,
      "grad_norm": 0.00020445660629775375,
      "learning_rate": 3.1238779174147223e-06,
      "loss": 0.0019,
      "step": 9400
    },
    {
      "epoch": 17.06,
      "grad_norm": 0.00033040347625501454,
      "learning_rate": 2.9443447037701977e-06,
      "loss": 0.002,
      "step": 9500
    },
    {
      "epoch": 17.24,
      "grad_norm": 0.00017785438103601336,
      "learning_rate": 2.7648114901256736e-06,
      "loss": 0.0011,
      "step": 9600
    },
    {
      "epoch": 17.41,
      "grad_norm": 0.0003500489692669362,
      "learning_rate": 2.585278276481149e-06,
      "loss": 0.023,
      "step": 9700
    },
    {
      "epoch": 17.59,
      "grad_norm": 0.0002393827890045941,
      "learning_rate": 2.405745062836625e-06,
      "loss": 0.0001,
      "step": 9800
    },
    {
      "epoch": 17.77,
      "grad_norm": 0.00016358286666218191,
      "learning_rate": 2.2262118491921008e-06,
      "loss": 0.0017,
      "step": 9900
    },
    {
      "epoch": 17.95,
      "grad_norm": 0.0005156854167580605,
      "learning_rate": 2.0466786355475767e-06,
      "loss": 0.0022,
      "step": 10000
    },
    {
      "epoch": 18.13,
      "grad_norm": 0.0001992402976611629,
      "learning_rate": 1.8671454219030521e-06,
      "loss": 0.006,
      "step": 10100
    },
    {
      "epoch": 18.31,
      "grad_norm": 0.00023926289577502757,
      "learning_rate": 1.687612208258528e-06,
      "loss": 0.0026,
      "step": 10200
    },
    {
      "epoch": 18.49,
      "grad_norm": 0.00018842410645447671,
      "learning_rate": 1.5080789946140036e-06,
      "loss": 0.0063,
      "step": 10300
    },
    {
      "epoch": 18.67,
      "grad_norm": 0.00017712044063955545,
      "learning_rate": 1.3285457809694793e-06,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 18.85,
      "grad_norm": 0.00030714995227754116,
      "learning_rate": 1.1490125673249552e-06,
      "loss": 0.0,
      "step": 10500
    },
    {
      "epoch": 19.03,
      "grad_norm": 0.0004120106459595263,
      "learning_rate": 9.694793536804308e-07,
      "loss": 0.0041,
      "step": 10600
    },
    {
      "epoch": 19.21,
      "grad_norm": 0.00010405273496871814,
      "learning_rate": 7.899461400359067e-07,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 19.39,
      "grad_norm": 9.223635424859822e-05,
      "learning_rate": 6.104129263913825e-07,
      "loss": 0.0001,
      "step": 10800
    },
    {
      "epoch": 19.57,
      "grad_norm": 0.00015765101124998182,
      "learning_rate": 4.3087971274685824e-07,
      "loss": 0.0042,
      "step": 10900
    },
    {
      "epoch": 19.75,
      "grad_norm": 0.00016278728435281664,
      "learning_rate": 2.5134649910233396e-07,
      "loss": 0.0003,
      "step": 11000
    },
    {
      "epoch": 19.93,
      "grad_norm": 0.00011305743100820109,
      "learning_rate": 7.18132854578097e-08,
      "loss": 0.0,
      "step": 11100
    }
  ],
  "logging_steps": 100,
  "max_steps": 11140,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 1.1371892211152928e+17,
  "train_batch_size": 14,
  "trial_name": null,
  "trial_params": null
}