qwen3-8b-dpo-avg-beta0p05-s42/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.6846090351065545,
    "train_runtime": 26441.528,
    "train_samples": 18644,
    "train_samples_per_second": 0.705,
    "train_steps_per_second": 0.011
}