qwen3-8b-ronpo-full-expect-s42/train_results.json

{
    "epoch": 0.06652991073903643,
    "total_flos": 0.0,
    "train_loss": 1.094520975748698,
    "train_runtime": 689.8738,
    "train_samples": 18037,
    "train_samples_per_second": 1.739,
    "train_steps_per_second": 0.109
}