{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 29658,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "learning_rate": 9.83309730932632e-06,
      "loss": 1.8178,
      "step": 500
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.66450873288826e-06,
      "loss": 1.5781,
      "step": 1000
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.4959201564502e-06,
      "loss": 1.5743,
      "step": 1500
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.327331580012139e-06,
      "loss": 1.5106,
      "step": 2000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.159080180726955e-06,
      "loss": 1.4435,
      "step": 2500
    },
    {
      "epoch": 0.61,
      "learning_rate": 8.990491604288895e-06,
      "loss": 1.4369,
      "step": 3000
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.821903027850834e-06,
      "loss": 1.4113,
      "step": 3500
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.653314451412773e-06,
      "loss": 1.3945,
      "step": 4000
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.484725874974711e-06,
      "loss": 1.4233,
      "step": 4500
    },
    {
      "epoch": 1.01,
      "learning_rate": 8.316137298536652e-06,
      "loss": 1.3204,
      "step": 5000
    },
    {
      "epoch": 1.11,
      "learning_rate": 8.147885899251467e-06,
      "loss": 1.1246,
      "step": 5500
    },
    {
      "epoch": 1.21,
      "learning_rate": 7.979297322813406e-06,
      "loss": 1.1869,
      "step": 6000
    },
    {
      "epoch": 1.31,
      "learning_rate": 7.810708746375346e-06,
      "loss": 1.149,
      "step": 6500
    },
    {
      "epoch": 1.42,
      "learning_rate": 7.642120169937287e-06,
      "loss": 1.1295,
      "step": 7000
    },
    {
      "epoch": 1.52,
      "learning_rate": 7.4735315934992245e-06,
      "loss": 1.1715,
      "step": 7500
    },
    {
      "epoch": 1.62,
      "learning_rate": 7.304943017061165e-06,
      "loss": 1.149,
      "step": 8000
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.136691617775981e-06,
      "loss": 1.1239,
      "step": 8500
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.968103041337919e-06,
      "loss": 1.1282,
      "step": 9000
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.799514464899859e-06,
      "loss": 1.1427,
      "step": 9500
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.630925888461799e-06,
      "loss": 1.0791,
      "step": 10000
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.462337312023738e-06,
      "loss": 0.9169,
      "step": 10500
    },
    {
      "epoch": 2.23,
      "learning_rate": 6.293748735585677e-06,
      "loss": 0.9361,
      "step": 11000
    },
    {
      "epoch": 2.33,
      "learning_rate": 6.125497336300493e-06,
      "loss": 0.9294,
      "step": 11500
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.956908759862432e-06,
      "loss": 0.9456,
      "step": 12000
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.7886573605772475e-06,
      "loss": 0.9395,
      "step": 12500
    },
    {
      "epoch": 2.63,
      "learning_rate": 5.620068784139188e-06,
      "loss": 0.9367,
      "step": 13000
    },
    {
      "epoch": 2.73,
      "learning_rate": 5.4514802077011266e-06,
      "loss": 0.9096,
      "step": 13500
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.282891631263066e-06,
      "loss": 0.9506,
      "step": 14000
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.114303054825005e-06,
      "loss": 0.9047,
      "step": 14500
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.945714478386945e-06,
      "loss": 0.8577,
      "step": 15000
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.777125901948884e-06,
      "loss": 0.7538,
      "step": 15500
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.608537325510824e-06,
      "loss": 0.8059,
      "step": 16000
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.439948749072764e-06,
      "loss": 0.7702,
      "step": 16500
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.271360172634702e-06,
      "loss": 0.7905,
      "step": 17000
    },
    {
      "epoch": 3.54,
      "learning_rate": 4.103108773349518e-06,
      "loss": 0.7902,
      "step": 17500
    },
    {
      "epoch": 3.64,
      "learning_rate": 3.934520196911458e-06,
      "loss": 0.7931,
      "step": 18000
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.765931620473397e-06,
      "loss": 0.7793,
      "step": 18500
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.5973430440353362e-06,
      "loss": 0.7803,
      "step": 19000
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.428754467597276e-06,
      "loss": 0.7636,
      "step": 19500
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.2605030683120915e-06,
      "loss": 0.6841,
      "step": 20000
    },
    {
      "epoch": 4.15,
      "learning_rate": 3.0919144918740306e-06,
      "loss": 0.6618,
      "step": 20500
    },
    {
      "epoch": 4.25,
      "learning_rate": 2.9233259154359705e-06,
      "loss": 0.6649,
      "step": 21000
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.75473733899791e-06,
      "loss": 0.6313,
      "step": 21500
    },
    {
      "epoch": 4.45,
      "learning_rate": 2.586148762559849e-06,
      "loss": 0.6625,
      "step": 22000
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.417897363274665e-06,
      "loss": 0.6405,
      "step": 22500
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.2493087868366044e-06,
      "loss": 0.7022,
      "step": 23000
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.0807202103985435e-06,
      "loss": 0.6901,
      "step": 23500
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.912131633960483e-06,
      "loss": 0.6647,
      "step": 24000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.7438802346752987e-06,
      "loss": 0.6386,
      "step": 24500
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.575291658237238e-06,
      "loss": 0.6375,
      "step": 25000
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.4067030817991776e-06,
      "loss": 0.5791,
      "step": 25500
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2381145053611169e-06,
      "loss": 0.5941,
      "step": 26000
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.0695259289230562e-06,
      "loss": 0.5801,
      "step": 26500
    },
    {
      "epoch": 5.46,
      "learning_rate": 9.009373524849957e-07,
      "loss": 0.5734,
      "step": 27000
    },
    {
      "epoch": 5.56,
      "learning_rate": 7.323487760469351e-07,
      "loss": 0.5798,
      "step": 27500
    },
    {
      "epoch": 5.66,
      "learning_rate": 5.640973767617506e-07,
      "loss": 0.5903,
      "step": 28000
    },
    {
      "epoch": 5.77,
      "learning_rate": 3.955088003236901e-07,
      "loss": 0.5849,
      "step": 28500
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.2692022388562955e-07,
      "loss": 0.5885,
      "step": 29000
    },
    {
      "epoch": 5.97,
      "learning_rate": 5.833164744756895e-08,
      "loss": 0.5763,
      "step": 29500
    }
  ],
  "logging_steps": 500,
  "max_steps": 29658,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 6.198533922642739e+16,
  "trial_name": null,
  "trial_params": null
}