{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.024089983172439,
  "global_step": 8500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07,
      "learning_rate": 9.92912827781715e-09,
      "loss": 2.0407,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.858256555634302e-09,
      "loss": 2.0254,
      "step": 200
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.787384833451452e-09,
      "loss": 2.0,
      "step": 300
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.716513111268604e-09,
      "loss": 2.021,
      "step": 400
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.645641389085754e-09,
      "loss": 2.021,
      "step": 500
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.574769666902906e-09,
      "loss": 2.0231,
      "step": 600
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.503897944720057e-09,
      "loss": 2.0069,
      "step": 700
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.433026222537208e-09,
      "loss": 2.012,
      "step": 800
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.362154500354359e-09,
      "loss": 2.0211,
      "step": 900
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.291282778171509e-09,
      "loss": 2.0124,
      "step": 1000
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.22041105598866e-09,
      "loss": 2.0022,
      "step": 1100
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.149539333805812e-09,
      "loss": 2.0183,
      "step": 1200
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.078667611622963e-09,
      "loss": 2.0216,
      "step": 1300
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.007795889440113e-09,
      "loss": 2.0148,
      "step": 1400
    },
    {
      "epoch": 1.06,
      "learning_rate": 8.936924167257265e-09,
      "loss": 2.037,
      "step": 1500
    },
    {
      "epoch": 1.13,
      "learning_rate": 8.866052445074415e-09,
      "loss": 2.0141,
      "step": 1600
    },
    {
      "epoch": 1.2,
      "learning_rate": 8.795180722891567e-09,
      "loss": 2.0392,
      "step": 1700
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.724309000708719e-09,
      "loss": 2.0102,
      "step": 1800
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.653437278525869e-09,
      "loss": 2.0282,
      "step": 1900
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.582565556343019e-09,
      "loss": 2.0227,
      "step": 2000
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.51169383416017e-09,
      "loss": 2.0205,
      "step": 2100
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.440822111977321e-09,
      "loss": 2.0152,
      "step": 2200
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.369950389794471e-09,
      "loss": 2.0001,
      "step": 2300
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.299078667611623e-09,
      "loss": 2.0182,
      "step": 2400
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.228206945428775e-09,
      "loss": 2.0066,
      "step": 2500
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.157335223245925e-09,
      "loss": 2.0176,
      "step": 2600
    },
    {
      "epoch": 1.91,
      "learning_rate": 8.086463501063075e-09,
      "loss": 2.0317,
      "step": 2700
    },
    {
      "epoch": 1.98,
      "learning_rate": 8.015591778880227e-09,
      "loss": 2.0088,
      "step": 2800
    },
    {
      "epoch": 2.06,
      "learning_rate": 7.944720056697377e-09,
      "loss": 2.0326,
      "step": 2900
    },
    {
      "epoch": 2.13,
      "learning_rate": 7.873848334514529e-09,
      "loss": 2.0085,
      "step": 3000
    },
    {
      "epoch": 2.2,
      "learning_rate": 7.802976612331681e-09,
      "loss": 2.0435,
      "step": 3100
    },
    {
      "epoch": 2.27,
      "learning_rate": 7.732104890148831e-09,
      "loss": 2.0367,
      "step": 3200
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.661233167965981e-09,
      "loss": 2.0237,
      "step": 3300
    },
    {
      "epoch": 2.41,
      "learning_rate": 7.590361445783133e-09,
      "loss": 2.0296,
      "step": 3400
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.519489723600283e-09,
      "loss": 2.0079,
      "step": 3500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.4486180014174344e-09,
      "loss": 2.0106,
      "step": 3600
    },
    {
      "epoch": 2.62,
      "learning_rate": 7.3777462792345854e-09,
      "loss": 2.0207,
      "step": 3700
    },
    {
      "epoch": 2.69,
      "learning_rate": 7.306874557051737e-09,
      "loss": 2.0215,
      "step": 3800
    },
    {
      "epoch": 2.76,
      "learning_rate": 7.2360028348688875e-09,
      "loss": 2.018,
      "step": 3900
    },
    {
      "epoch": 2.83,
      "learning_rate": 7.1651311126860385e-09,
      "loss": 1.996,
      "step": 4000
    },
    {
      "epoch": 2.91,
      "learning_rate": 7.0942593905031895e-09,
      "loss": 2.0043,
      "step": 4100
    },
    {
      "epoch": 2.98,
      "learning_rate": 7.0233876683203405e-09,
      "loss": 2.0007,
      "step": 4200
    },
    {
      "epoch": 3.05,
      "learning_rate": 6.952515946137491e-09,
      "loss": 2.0034,
      "step": 4300
    },
    {
      "epoch": 3.12,
      "learning_rate": 6.881644223954642e-09,
      "loss": 2.0066,
      "step": 4400
    },
    {
      "epoch": 3.19,
      "learning_rate": 6.810772501771794e-09,
      "loss": 1.9741,
      "step": 4500
    },
    {
      "epoch": 3.26,
      "learning_rate": 6.739900779588945e-09,
      "loss": 1.9767,
      "step": 4600
    },
    {
      "epoch": 3.33,
      "learning_rate": 6.669029057406096e-09,
      "loss": 1.9898,
      "step": 4700
    },
    {
      "epoch": 3.4,
      "learning_rate": 6.598157335223246e-09,
      "loss": 1.9723,
      "step": 4800
    },
    {
      "epoch": 3.47,
      "learning_rate": 6.527285613040397e-09,
      "loss": 1.9864,
      "step": 4900
    },
    {
      "epoch": 3.54,
      "learning_rate": 6.456413890857548e-09,
      "loss": 1.9762,
      "step": 5000
    },
    {
      "epoch": 3.61,
      "learning_rate": 6.385542168674698e-09,
      "loss": 2.0763,
      "step": 5100
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.31467044649185e-09,
      "loss": 2.065,
      "step": 5200
    },
    {
      "epoch": 3.76,
      "learning_rate": 6.243798724309001e-09,
      "loss": 2.06,
      "step": 5300
    },
    {
      "epoch": 3.83,
      "learning_rate": 6.172927002126152e-09,
      "loss": 2.0403,
      "step": 5400
    },
    {
      "epoch": 3.9,
      "learning_rate": 6.102055279943303e-09,
      "loss": 2.0715,
      "step": 5500
    },
    {
      "epoch": 3.97,
      "learning_rate": 6.031183557760453e-09,
      "loss": 2.0762,
      "step": 5600
    },
    {
      "epoch": 4.04,
      "learning_rate": 5.960311835577604e-09,
      "loss": 2.0365,
      "step": 5700
    },
    {
      "epoch": 4.11,
      "learning_rate": 5.889440113394756e-09,
      "loss": 2.0154,
      "step": 5800
    },
    {
      "epoch": 4.18,
      "learning_rate": 5.818568391211907e-09,
      "loss": 2.0166,
      "step": 5900
    },
    {
      "epoch": 4.25,
      "learning_rate": 5.747696669029058e-09,
      "loss": 2.0339,
      "step": 6000
    },
    {
      "epoch": 4.32,
      "learning_rate": 5.676824946846208e-09,
      "loss": 2.0451,
      "step": 6100
    },
    {
      "epoch": 4.39,
      "learning_rate": 5.605953224663359e-09,
      "loss": 2.023,
      "step": 6200
    },
    {
      "epoch": 4.46,
      "learning_rate": 5.53508150248051e-09,
      "loss": 2.0222,
      "step": 6300
    },
    {
      "epoch": 4.54,
      "learning_rate": 5.464209780297661e-09,
      "loss": 2.0037,
      "step": 6400
    },
    {
      "epoch": 4.61,
      "learning_rate": 5.393338058114813e-09,
      "loss": 2.0116,
      "step": 6500
    },
    {
      "epoch": 4.68,
      "learning_rate": 5.322466335931963e-09,
      "loss": 2.0236,
      "step": 6600
    },
    {
      "epoch": 4.75,
      "learning_rate": 5.251594613749114e-09,
      "loss": 2.0302,
      "step": 6700
    },
    {
      "epoch": 4.82,
      "learning_rate": 5.180722891566265e-09,
      "loss": 2.0236,
      "step": 6800
    },
    {
      "epoch": 4.89,
      "learning_rate": 5.109851169383416e-09,
      "loss": 2.015,
      "step": 6900
    },
    {
      "epoch": 4.96,
      "learning_rate": 5.038979447200567e-09,
      "loss": 2.0037,
      "step": 7000
    },
    {
      "epoch": 5.03,
      "learning_rate": 4.9681077250177184e-09,
      "loss": 2.0319,
      "step": 7100
    },
    {
      "epoch": 5.1,
      "learning_rate": 4.897236002834869e-09,
      "loss": 2.0271,
      "step": 7200
    },
    {
      "epoch": 5.17,
      "learning_rate": 4.82636428065202e-09,
      "loss": 2.0337,
      "step": 7300
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.755492558469171e-09,
      "loss": 2.0305,
      "step": 7400
    },
    {
      "epoch": 5.32,
      "learning_rate": 4.684620836286322e-09,
      "loss": 2.0261,
      "step": 7500
    },
    {
      "epoch": 5.39,
      "learning_rate": 4.613749114103473e-09,
      "loss": 2.0234,
      "step": 7600
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.542877391920624e-09,
      "loss": 2.0209,
      "step": 7700
    },
    {
      "epoch": 5.53,
      "learning_rate": 4.472005669737775e-09,
      "loss": 2.0166,
      "step": 7800
    },
    {
      "epoch": 5.6,
      "learning_rate": 4.401133947554926e-09,
      "loss": 2.0268,
      "step": 7900
    },
    {
      "epoch": 5.67,
      "learning_rate": 4.330262225372077e-09,
      "loss": 2.0199,
      "step": 8000
    },
    {
      "epoch": 5.74,
      "learning_rate": 4.259390503189228e-09,
      "loss": 2.005,
      "step": 8100
    },
    {
      "epoch": 5.81,
      "learning_rate": 4.188518781006379e-09,
      "loss": 2.0096,
      "step": 8200
    },
    {
      "epoch": 5.88,
      "learning_rate": 4.117647058823529e-09,
      "loss": 2.0239,
      "step": 8300
    },
    {
      "epoch": 5.95,
      "learning_rate": 4.046775336640681e-09,
      "loss": 2.0096,
      "step": 8400
    },
    {
      "epoch": 6.02,
      "learning_rate": 3.975903614457831e-09,
      "loss": 2.0432,
      "step": 8500
    }
  ],
  "max_steps": 14110,
  "num_train_epochs": 10,
  "total_flos": 6.087535537968384e+17,
  "trial_name": null,
  "trial_params": null
}