train_results.json

{
    "effective_tokens_per_sec": 2777.966393725024,
    "epoch": 3.0,
    "total_flos": 2.510120369642275e+18,
    "train_loss": 1.2744095386493774,
    "train_runtime": 14979.881,
    "train_samples_per_second": 1.794,
    "train_steps_per_second": 0.112
}