train_results.json

{
    "epoch": 4.887459807073955,
    "num_input_tokens_seen": 1208400,
    "total_flos": 5.441370708980531e+16,
    "train_loss": 0.5078434096239538,
    "train_runtime": 2562.642,
    "train_samples_per_second": 9.693,
    "train_steps_per_second": 0.074
}