{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 81250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.6109,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.5501,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.5e-06,
      "loss": 1.1228,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.8543,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.5e-06,
      "loss": 0.8183,
      "step": 2500
    },
    {
      "epoch": 0.07,
      "learning_rate": 3e-06,
      "loss": 0.7931,
      "step": 3000
    },
    {
      "epoch": 0.09,
      "learning_rate": 3.5e-06,
      "loss": 0.7792,
      "step": 3500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7713,
      "step": 4000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.5e-06,
      "loss": 0.7678,
      "step": 4500
    },
    {
      "epoch": 0.12,
      "learning_rate": 5e-06,
      "loss": 0.7736,
      "step": 5000
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.7527,
      "step": 5500
    },
    {
      "epoch": 0.15,
      "learning_rate": 6e-06,
      "loss": 0.7499,
      "step": 6000
    },
    {
      "epoch": 0.16,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.7438,
      "step": 6500
    },
    {
      "epoch": 0.17,
      "learning_rate": 7e-06,
      "loss": 0.7462,
      "step": 7000
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.7524,
      "step": 7500
    },
    {
      "epoch": 0.2,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.7427,
      "step": 8000
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.5e-06,
      "loss": 0.7342,
      "step": 8500
    },
    {
      "epoch": 0.22,
      "learning_rate": 9e-06,
      "loss": 0.7347,
      "step": 9000
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.5e-06,
      "loss": 0.749,
      "step": 9500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.7288,
      "step": 10000
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.993769470404985e-06,
      "loss": 0.7326,
      "step": 10500
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.98753894080997e-06,
      "loss": 0.7256,
      "step": 11000
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.981308411214954e-06,
      "loss": 0.7386,
      "step": 11500
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.97507788161994e-06,
      "loss": 0.7285,
      "step": 12000
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.968847352024923e-06,
      "loss": 0.7223,
      "step": 12500
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.962616822429907e-06,
      "loss": 0.7327,
      "step": 13000
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.956386292834892e-06,
      "loss": 0.7189,
      "step": 13500
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.950155763239876e-06,
      "loss": 0.7316,
      "step": 14000
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.943925233644862e-06,
      "loss": 0.7129,
      "step": 14500
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.937694704049845e-06,
      "loss": 0.7096,
      "step": 15000
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.931464174454829e-06,
      "loss": 0.7134,
      "step": 15500
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.925233644859815e-06,
      "loss": 0.7201,
      "step": 16000
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.919003115264798e-06,
      "loss": 0.7185,
      "step": 16500
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.912772585669784e-06,
      "loss": 0.7215,
      "step": 17000
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.906542056074768e-06,
      "loss": 0.7142,
      "step": 17500
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.900311526479751e-06,
      "loss": 0.7261,
      "step": 18000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.894080996884737e-06,
      "loss": 0.7091,
      "step": 18500
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.88785046728972e-06,
      "loss": 0.7138,
      "step": 19000
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.881619937694706e-06,
      "loss": 0.7134,
      "step": 19500
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.87538940809969e-06,
      "loss": 0.7157,
      "step": 20000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.869158878504674e-06,
      "loss": 0.7136,
      "step": 20500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.862928348909659e-06,
      "loss": 0.7133,
      "step": 21000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.856697819314643e-06,
      "loss": 0.7048,
      "step": 21500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.850467289719627e-06,
      "loss": 0.7004,
      "step": 22000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.844236760124612e-06,
      "loss": 0.69,
      "step": 22500
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.838006230529596e-06,
      "loss": 0.7158,
      "step": 23000
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.83177570093458e-06,
      "loss": 0.6995,
      "step": 23500
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.825545171339565e-06,
      "loss": 0.6893,
      "step": 24000
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.819314641744549e-06,
      "loss": 0.6978,
      "step": 24500
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.813084112149533e-06,
      "loss": 0.7025,
      "step": 25000
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.806853582554518e-06,
      "loss": 0.6952,
      "step": 25500
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.800623052959502e-06,
      "loss": 0.7102,
      "step": 26000
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.794392523364486e-06,
      "loss": 0.7042,
      "step": 26500
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.788161993769471e-06,
      "loss": 0.6953,
      "step": 27000
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.781931464174455e-06,
      "loss": 0.7247,
      "step": 27500
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.775700934579439e-06,
      "loss": 0.7081,
      "step": 28000
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.769470404984424e-06,
      "loss": 0.711,
      "step": 28500
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.763239875389408e-06,
      "loss": 0.6943,
      "step": 29000
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.757009345794393e-06,
      "loss": 0.6845,
      "step": 29500
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.750778816199377e-06,
      "loss": 0.7012,
      "step": 30000
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.744548286604361e-06,
      "loss": 0.7031,
      "step": 30500
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.738317757009347e-06,
      "loss": 0.6994,
      "step": 31000
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.73208722741433e-06,
      "loss": 0.7058,
      "step": 31500
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.725856697819316e-06,
      "loss": 0.69,
      "step": 32000
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.7196261682243e-06,
      "loss": 0.6888,
      "step": 32500
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.713395638629283e-06,
      "loss": 0.7044,
      "step": 33000
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.707165109034269e-06,
      "loss": 0.6841,
      "step": 33500
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.700934579439253e-06,
      "loss": 0.6886,
      "step": 34000
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.694704049844238e-06,
      "loss": 0.6884,
      "step": 34500
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.688473520249222e-06,
      "loss": 0.699,
      "step": 35000
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.682242990654206e-06,
      "loss": 0.7042,
      "step": 35500
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.676012461059191e-06,
      "loss": 0.6993,
      "step": 36000
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.669781931464175e-06,
      "loss": 0.6983,
      "step": 36500
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.66355140186916e-06,
      "loss": 0.697,
      "step": 37000
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.657320872274144e-06,
      "loss": 0.704,
      "step": 37500
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.651090342679128e-06,
      "loss": 0.6917,
      "step": 38000
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.644859813084113e-06,
      "loss": 0.6904,
      "step": 38500
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.638629283489097e-06,
      "loss": 0.6946,
      "step": 39000
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.632398753894083e-06,
      "loss": 0.6905,
      "step": 39500
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.626168224299066e-06,
      "loss": 0.7,
      "step": 40000
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.61993769470405e-06,
      "loss": 0.7068,
      "step": 40500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.69898,
      "eval_loss": 0.6849758625030518,
      "eval_runtime": 138.1081,
      "eval_samples_per_second": 362.035,
      "eval_steps_per_second": 22.627,
      "step": 40625
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.613707165109036e-06,
      "loss": 0.6541,
      "step": 41000
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.60747663551402e-06,
      "loss": 0.6604,
      "step": 41500
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.601246105919005e-06,
      "loss": 0.6514,
      "step": 42000
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.595015576323989e-06,
      "loss": 0.6295,
      "step": 42500
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.588785046728972e-06,
      "loss": 0.6594,
      "step": 43000
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.582554517133958e-06,
      "loss": 0.6389,
      "step": 43500
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.576323987538942e-06,
      "loss": 0.6329,
      "step": 44000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.570093457943927e-06,
      "loss": 0.6528,
      "step": 44500
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.563862928348911e-06,
      "loss": 0.6429,
      "step": 45000
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.557632398753895e-06,
      "loss": 0.6358,
      "step": 45500
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.55140186915888e-06,
      "loss": 0.6328,
      "step": 46000
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.545171339563864e-06,
      "loss": 0.6362,
      "step": 46500
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.538940809968848e-06,
      "loss": 0.6299,
      "step": 47000
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.532710280373833e-06,
      "loss": 0.6458,
      "step": 47500
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.526479750778817e-06,
      "loss": 0.6404,
      "step": 48000
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.520249221183802e-06,
      "loss": 0.6451,
      "step": 48500
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.514018691588786e-06,
      "loss": 0.6299,
      "step": 49000
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.50778816199377e-06,
      "loss": 0.6411,
      "step": 49500
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.501557632398755e-06,
      "loss": 0.6398,
      "step": 50000
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.49532710280374e-06,
      "loss": 0.6471,
      "step": 50500
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.489096573208723e-06,
      "loss": 0.6479,
      "step": 51000
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.482866043613708e-06,
      "loss": 0.6239,
      "step": 51500
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.476635514018692e-06,
      "loss": 0.6405,
      "step": 52000
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.470404984423676e-06,
      "loss": 0.6458,
      "step": 52500
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.464174454828661e-06,
      "loss": 0.6269,
      "step": 53000
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.457943925233645e-06,
      "loss": 0.6438,
      "step": 53500
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.451713395638629e-06,
      "loss": 0.65,
      "step": 54000
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.445482866043614e-06,
      "loss": 0.6495,
      "step": 54500
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.439252336448598e-06,
      "loss": 0.6355,
      "step": 55000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.433021806853582e-06,
      "loss": 0.6452,
      "step": 55500
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.426791277258568e-06,
      "loss": 0.6419,
      "step": 56000
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.420560747663551e-06,
      "loss": 0.6589,
      "step": 56500
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.414330218068537e-06,
      "loss": 0.6426,
      "step": 57000
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.40809968847352e-06,
      "loss": 0.6367,
      "step": 57500
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.401869158878504e-06,
      "loss": 0.6388,
      "step": 58000
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.39563862928349e-06,
      "loss": 0.6507,
      "step": 58500
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.389408099688474e-06,
      "loss": 0.6377,
      "step": 59000
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.383177570093459e-06,
      "loss": 0.6612,
      "step": 59500
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.376947040498443e-06,
      "loss": 0.659,
      "step": 60000
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.370716510903427e-06,
      "loss": 0.6396,
      "step": 60500
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.364485981308412e-06,
      "loss": 0.6541,
      "step": 61000
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.358255451713396e-06,
      "loss": 0.6483,
      "step": 61500
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.352024922118381e-06,
      "loss": 0.6366,
      "step": 62000
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.345794392523365e-06,
      "loss": 0.6476,
      "step": 62500
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.339563862928349e-06,
      "loss": 0.6553,
      "step": 63000
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.6362,
      "step": 63500
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.327102803738318e-06,
      "loss": 0.6442,
      "step": 64000
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.320872274143304e-06,
      "loss": 0.6566,
      "step": 64500
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.314641744548287e-06,
      "loss": 0.6497,
      "step": 65000
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.308411214953271e-06,
      "loss": 0.6461,
      "step": 65500
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.302180685358257e-06,
      "loss": 0.6609,
      "step": 66000
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.29595015576324e-06,
      "loss": 0.6502,
      "step": 66500
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.289719626168226e-06,
      "loss": 0.6531,
      "step": 67000
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.28348909657321e-06,
      "loss": 0.6441,
      "step": 67500
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.277258566978193e-06,
      "loss": 0.6487,
      "step": 68000
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.271028037383179e-06,
      "loss": 0.6436,
      "step": 68500
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.264797507788163e-06,
      "loss": 0.6575,
      "step": 69000
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.258566978193148e-06,
      "loss": 0.6514,
      "step": 69500
    },
    {
      "epoch": 1.72,
      "learning_rate": 9.252336448598132e-06,
      "loss": 0.6424,
      "step": 70000
    },
    {
      "epoch": 1.74,
      "learning_rate": 9.246105919003116e-06,
      "loss": 0.646,
      "step": 70500
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.239875389408101e-06,
      "loss": 0.6445,
      "step": 71000
    },
    {
      "epoch": 1.76,
      "learning_rate": 9.233644859813085e-06,
      "loss": 0.6507,
      "step": 71500
    },
    {
      "epoch": 1.77,
      "learning_rate": 9.22741433021807e-06,
      "loss": 0.6564,
      "step": 72000
    },
    {
      "epoch": 1.78,
      "learning_rate": 9.221183800623054e-06,
      "loss": 0.6422,
      "step": 72500
    },
    {
      "epoch": 1.8,
      "learning_rate": 9.214953271028038e-06,
      "loss": 0.6408,
      "step": 73000
    },
    {
      "epoch": 1.81,
      "learning_rate": 9.208722741433023e-06,
      "loss": 0.6394,
      "step": 73500
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.202492211838007e-06,
      "loss": 0.6421,
      "step": 74000
    },
    {
      "epoch": 1.83,
      "learning_rate": 9.196261682242991e-06,
      "loss": 0.6379,
      "step": 74500
    },
    {
      "epoch": 1.85,
      "learning_rate": 9.190031152647976e-06,
      "loss": 0.6421,
      "step": 75000
    },
    {
      "epoch": 1.86,
      "learning_rate": 9.18380062305296e-06,
      "loss": 0.6277,
      "step": 75500
    },
    {
      "epoch": 1.87,
      "learning_rate": 9.177570093457944e-06,
      "loss": 0.6475,
      "step": 76000
    },
    {
      "epoch": 1.88,
      "learning_rate": 9.17133956386293e-06,
      "loss": 0.65,
      "step": 76500
    },
    {
      "epoch": 1.9,
      "learning_rate": 9.165109034267913e-06,
      "loss": 0.6546,
      "step": 77000
    },
    {
      "epoch": 1.91,
      "learning_rate": 9.158878504672899e-06,
      "loss": 0.6476,
      "step": 77500
    },
    {
      "epoch": 1.92,
      "learning_rate": 9.152647975077882e-06,
      "loss": 0.6362,
      "step": 78000
    },
    {
      "epoch": 1.93,
      "learning_rate": 9.146417445482866e-06,
      "loss": 0.6368,
      "step": 78500
    },
    {
      "epoch": 1.94,
      "learning_rate": 9.140186915887852e-06,
      "loss": 0.6338,
      "step": 79000
    },
    {
      "epoch": 1.96,
      "learning_rate": 9.133956386292835e-06,
      "loss": 0.6456,
      "step": 79500
    },
    {
      "epoch": 1.97,
      "learning_rate": 9.12772585669782e-06,
      "loss": 0.6563,
      "step": 80000
    },
    {
      "epoch": 1.98,
      "learning_rate": 9.121495327102805e-06,
      "loss": 0.6456,
      "step": 80500
    },
    {
      "epoch": 1.99,
      "learning_rate": 9.115264797507789e-06,
      "loss": 0.6518,
      "step": 81000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.70374,
      "eval_loss": 0.679389238357544,
      "eval_runtime": 138.0175,
      "eval_samples_per_second": 362.273,
      "eval_steps_per_second": 22.642,
      "step": 81250
    }
  ],
  "max_steps": 812500,
  "num_train_epochs": 20,
  "total_flos": 1.710267926016e+17,
  "trial_name": null,
  "trial_params": null
}