qwen3-8b-aaai27-flagship-ronpo-full-expect-s43/train_results.json

{
    "epoch": 0.2860548271752086,
    "total_flos": 0.0,
    "train_loss": 0.48261189058423043,
    "train_runtime": 8509.445,
    "train_samples": 50340,
    "train_samples_per_second": 1.692,
    "train_steps_per_second": 0.106
}