train_results.json

{
    "epoch": 4.951768488745981,
    "num_input_tokens_seen": 5192736,
    "total_flos": 2.3382655808988774e+17,
    "train_loss": 0.7082552919200585,
    "train_runtime": 5092.4527,
    "train_samples_per_second": 19.519,
    "train_steps_per_second": 0.076
}