Qwen3-4B-grpo-medmcqa/train_results.json

{
    "total_flos": 0.0,
    "train_loss": 0.0036022694108365965,
    "train_runtime": 15331.4788,
    "train_samples": 182822,
    "train_samples_per_second": 5.962,
    "train_steps_per_second": 0.124
}