train_results.json

{
    "effective_tokens_per_sec": 2865.9471019813323,
    "epoch": 3.0,
    "total_flos": 3.6341125364349665e+18,
    "train_loss": 1.4411861958004843,
    "train_runtime": 21029.937,
    "train_samples_per_second": 1.563,
    "train_steps_per_second": 0.098
}