Qwen2.5-3B-MATH-GRPO/all_results.json

{
    "total_flos": 0.0,
    "train_loss": 5.459580092858046e-10,
    "train_runtime": 31647.8785,
    "train_samples": 7493,
    "train_samples_per_second": 1.184,
    "train_steps_per_second": 0.037
}