train_results.json

{
    "effective_tokens_per_sec": 6441.624368872594,
    "epoch": 3.0,
    "total_flos": 4.4194007019526554e+18,
    "train_loss": 1.0514326345836018,
    "train_runtime": 11229.2105,
    "train_samples_per_second": 4.014,
    "train_steps_per_second": 0.126
}