llama-3-8b-base-new-dpo-hh-harmless-s_star0.6-4xh200-batch-64-20260421-213851/train_results.json

{
    "epoch": 0.999244142101285,
    "total_flos": 0.0,
    "train_loss": 1.103111317586971,
    "train_runtime": 1749.744,
    "train_samples": 42336,
    "train_samples_per_second": 24.196,
    "train_steps_per_second": 0.378
}