{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 70827,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.964702726361416e-05,
      "loss": 2.9121,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.929405452722832e-05,
      "loss": 2.1182,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.894108179084248e-05,
      "loss": 1.8923,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.858810905445664e-05,
      "loss": 1.7635,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.823513631807079e-05,
      "loss": 1.6796,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7882163581684954e-05,
      "loss": 1.5869,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752919084529911e-05,
      "loss": 1.5434,
      "step": 3500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7176218108913265e-05,
      "loss": 1.5104,
      "step": 4000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.682324537252743e-05,
      "loss": 1.4742,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.647027263614159e-05,
      "loss": 1.4164,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6117299899755744e-05,
      "loss": 1.3891,
      "step": 5500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.57643271633699e-05,
      "loss": 1.3882,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.541135442698406e-05,
      "loss": 1.3588,
      "step": 6500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5058381690598224e-05,
      "loss": 1.324,
      "step": 7000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.470540895421238e-05,
      "loss": 1.3183,
      "step": 7500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.4352436217826534e-05,
      "loss": 1.2891,
      "step": 8000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.3999463481440696e-05,
      "loss": 1.3059,
      "step": 8500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.364649074505485e-05,
      "loss": 1.2529,
      "step": 9000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3293518008669014e-05,
      "loss": 1.2591,
      "step": 9500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.2940545272283176e-05,
      "loss": 1.2502,
      "step": 10000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.258757253589733e-05,
      "loss": 1.2253,
      "step": 10500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2234599799511486e-05,
      "loss": 1.2159,
      "step": 11000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.188162706312564e-05,
      "loss": 1.2184,
      "step": 11500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1528654326739804e-05,
      "loss": 1.1937,
      "step": 12000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1175681590353966e-05,
      "loss": 1.1942,
      "step": 12500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.082270885396812e-05,
      "loss": 1.1704,
      "step": 13000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.046973611758228e-05,
      "loss": 1.1654,
      "step": 13500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.011676338119644e-05,
      "loss": 1.1603,
      "step": 14000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9763790644810593e-05,
      "loss": 1.1674,
      "step": 14500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9410817908424756e-05,
      "loss": 1.1685,
      "step": 15000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.905784517203892e-05,
      "loss": 1.1318,
      "step": 15500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.870487243565307e-05,
      "loss": 1.1255,
      "step": 16000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.835189969926723e-05,
      "loss": 1.1105,
      "step": 16500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.799892696288139e-05,
      "loss": 1.1255,
      "step": 17000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.764595422649555e-05,
      "loss": 1.1111,
      "step": 17500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.729298149010971e-05,
      "loss": 1.1029,
      "step": 18000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.694000875372386e-05,
      "loss": 1.1211,
      "step": 18500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6587036017338025e-05,
      "loss": 1.0865,
      "step": 19000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.623406328095218e-05,
      "loss": 1.0844,
      "step": 19500
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5881090544566335e-05,
      "loss": 1.0869,
      "step": 20000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.55281178081805e-05,
      "loss": 1.0624,
      "step": 20500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.517514507179466e-05,
      "loss": 1.0714,
      "step": 21000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.4822172335408815e-05,
      "loss": 1.0743,
      "step": 21500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.446919959902297e-05,
      "loss": 1.0653,
      "step": 22000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.411622686263713e-05,
      "loss": 1.0492,
      "step": 22500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.3763254126251294e-05,
      "loss": 1.0448,
      "step": 23000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.341028138986545e-05,
      "loss": 1.0286,
      "step": 23500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.3057308653479605e-05,
      "loss": 0.9574,
      "step": 24000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.270433591709377e-05,
      "loss": 0.9522,
      "step": 24500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.235136318070792e-05,
      "loss": 0.9199,
      "step": 25000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.199839044432208e-05,
      "loss": 0.9462,
      "step": 25500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.164541770793624e-05,
      "loss": 0.9239,
      "step": 26000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.12924449715504e-05,
      "loss": 0.9232,
      "step": 26500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.093947223516456e-05,
      "loss": 0.9096,
      "step": 27000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.058649949877871e-05,
      "loss": 0.9352,
      "step": 27500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.023352676239287e-05,
      "loss": 0.9241,
      "step": 28000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9880554026007036e-05,
      "loss": 0.9374,
      "step": 28500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.952758128962119e-05,
      "loss": 0.922,
      "step": 29000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.917460855323535e-05,
      "loss": 0.9204,
      "step": 29500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.882163581684951e-05,
      "loss": 0.9226,
      "step": 30000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8468663080463664e-05,
      "loss": 0.9082,
      "step": 30500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8115690344077826e-05,
      "loss": 0.9114,
      "step": 31000
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7762717607691985e-05,
      "loss": 0.9124,
      "step": 31500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7409744871306143e-05,
      "loss": 0.9146,
      "step": 32000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.70567721349203e-05,
      "loss": 0.9126,
      "step": 32500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6703799398534457e-05,
      "loss": 0.9029,
      "step": 33000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.6350826662148616e-05,
      "loss": 0.9048,
      "step": 33500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.5997853925762778e-05,
      "loss": 0.8891,
      "step": 34000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5644881189376933e-05,
      "loss": 0.9128,
      "step": 34500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5291908452991092e-05,
      "loss": 0.8802,
      "step": 35000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.493893571660525e-05,
      "loss": 0.8811,
      "step": 35500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.458596298021941e-05,
      "loss": 0.9011,
      "step": 36000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4232990243833568e-05,
      "loss": 0.8747,
      "step": 36500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3880017507447727e-05,
      "loss": 0.89,
      "step": 37000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3527044771061885e-05,
      "loss": 0.8768,
      "step": 37500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.317407203467604e-05,
      "loss": 0.8892,
      "step": 38000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2821099298290203e-05,
      "loss": 0.889,
      "step": 38500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.246812656190436e-05,
      "loss": 0.8801,
      "step": 39000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2115153825518516e-05,
      "loss": 0.8927,
      "step": 39500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.176218108913268e-05,
      "loss": 0.8796,
      "step": 40000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1409208352746834e-05,
      "loss": 0.8626,
      "step": 40500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1056235616360992e-05,
      "loss": 0.86,
      "step": 41000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.070326287997515e-05,
      "loss": 0.8748,
      "step": 41500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.035029014358931e-05,
      "loss": 0.8815,
      "step": 42000
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.999731740720347e-05,
      "loss": 0.8582,
      "step": 42500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9644344670817627e-05,
      "loss": 0.8764,
      "step": 43000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9291371934431786e-05,
      "loss": 0.8639,
      "step": 43500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8938399198045944e-05,
      "loss": 0.86,
      "step": 44000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8585426461660103e-05,
      "loss": 0.8554,
      "step": 44500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.823245372527426e-05,
      "loss": 0.8509,
      "step": 45000
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.787948098888842e-05,
      "loss": 0.8504,
      "step": 45500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7526508252502576e-05,
      "loss": 0.8626,
      "step": 46000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7173535516116738e-05,
      "loss": 0.8619,
      "step": 46500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6820562779730893e-05,
      "loss": 0.8309,
      "step": 47000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.646759004334505e-05,
      "loss": 0.7807,
      "step": 47500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6114617306959214e-05,
      "loss": 0.728,
      "step": 48000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.576164457057337e-05,
      "loss": 0.7391,
      "step": 48500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5408671834187528e-05,
      "loss": 0.7315,
      "step": 49000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5055699097801686e-05,
      "loss": 0.7278,
      "step": 49500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4702726361415845e-05,
      "loss": 0.7339,
      "step": 50000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4349753625030005e-05,
      "loss": 0.728,
      "step": 50500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3996780888644162e-05,
      "loss": 0.7272,
      "step": 51000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.364380815225832e-05,
      "loss": 0.7108,
      "step": 51500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.329083541587248e-05,
      "loss": 0.732,
      "step": 52000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.2937862679486637e-05,
      "loss": 0.7358,
      "step": 52500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2584889943100794e-05,
      "loss": 0.707,
      "step": 53000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2231917206714954e-05,
      "loss": 0.7424,
      "step": 53500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1878944470329113e-05,
      "loss": 0.7195,
      "step": 54000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1525971733943271e-05,
      "loss": 0.7217,
      "step": 54500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.117299899755743e-05,
      "loss": 0.7395,
      "step": 55000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0820026261171589e-05,
      "loss": 0.7181,
      "step": 55500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0467053524785746e-05,
      "loss": 0.7017,
      "step": 56000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0114080788399904e-05,
      "loss": 0.7223,
      "step": 56500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.761108052014063e-06,
      "loss": 0.7163,
      "step": 57000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.40813531562822e-06,
      "loss": 0.7108,
      "step": 57500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.05516257924238e-06,
      "loss": 0.7059,
      "step": 58000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.702189842856539e-06,
      "loss": 0.7209,
      "step": 58500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.349217106470698e-06,
      "loss": 0.7192,
      "step": 59000
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.996244370084854e-06,
      "loss": 0.7092,
      "step": 59500
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.643271633699013e-06,
      "loss": 0.7138,
      "step": 60000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.290298897313172e-06,
      "loss": 0.7356,
      "step": 60500
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.9373261609273305e-06,
      "loss": 0.7045,
      "step": 61000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.584353424541488e-06,
      "loss": 0.7004,
      "step": 61500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.231380688155647e-06,
      "loss": 0.7035,
      "step": 62000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.878407951769806e-06,
      "loss": 0.7067,
      "step": 62500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.5254352153839634e-06,
      "loss": 0.7039,
      "step": 63000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.172462478998123e-06,
      "loss": 0.6999,
      "step": 63500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.819489742612281e-06,
      "loss": 0.6821,
      "step": 64000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.4665170062264394e-06,
      "loss": 0.7024,
      "step": 64500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.113544269840597e-06,
      "loss": 0.6987,
      "step": 65000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.7605715334547563e-06,
      "loss": 0.7061,
      "step": 65500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.4075987970689146e-06,
      "loss": 0.6942,
      "step": 66000
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.054626060683073e-06,
      "loss": 0.7083,
      "step": 66500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.701653324297231e-06,
      "loss": 0.6986,
      "step": 67000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3486805879113897e-06,
      "loss": 0.6865,
      "step": 67500
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.995707851525548e-06,
      "loss": 0.705,
      "step": 68000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6427351151397066e-06,
      "loss": 0.7014,
      "step": 68500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.289762378753865e-06,
      "loss": 0.7101,
      "step": 69000
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.367896423680236e-07,
      "loss": 0.683,
      "step": 69500
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.83816905982182e-07,
      "loss": 0.6975,
      "step": 70000
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.308441695963404e-07,
      "loss": 0.6939,
      "step": 70500
    },
    {
      "epoch": 3.0,
      "step": 70827,
      "total_flos": 8.056348742334874e+16,
      "train_loss": 0.9704105395994184,
      "train_runtime": 18798.0807,
      "train_samples_per_second": 37.676,
      "train_steps_per_second": 3.768
    }
  ],
  "max_steps": 70827,
  "num_train_epochs": 3,
  "total_flos": 8.056348742334874e+16,
  "trial_name": null,
  "trial_params": null
}