qwen3-8b-dpo-avg-beta0p01-s42/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.6896600347675689,
    "train_runtime": 29629.3533,
    "train_samples": 18644,
    "train_samples_per_second": 0.629,
    "train_steps_per_second": 0.01
}