all_results.json

{
    "entropy": 2.5530850489934287,
    "epoch": 2.0,
    "eval_entropy": 2.5582991441090903,
    "eval_loss": 0.4179121255874634,
    "eval_mean_token_accuracy": 0.9070041179656982,
    "eval_num_tokens": 620664491.0,
    "eval_runtime": 0.4825,
    "eval_samples": 1000,
    "eval_samples_per_second": 373.052,
    "eval_steps_per_second": 6.218,
    "mean_token_accuracy": 0.902903805176417,
    "num_tokens": 620664491.0,
    "total_flos": 3.623983417768542e+18,
    "train_loss": 0.550684680861008,
    "train_runtime": 5178.5318,
    "train_samples": 1859300,
    "train_samples_per_second": 118.293,
    "train_steps_per_second": 0.924,
    "weight_norm": 725.9066744423832
}