train_results.json

{
    "effective_tokens_per_sec": 2864.090238424537,
    "epoch": 3.0,
    "total_flos": 2.4882019145802056e+18,
    "train_loss": 1.4038474378721917,
    "train_runtime": 17010.8289,
    "train_samples_per_second": 1.845,
    "train_steps_per_second": 0.115
}