Qwen-2.5-3B-GRPO-Math / train_results.json
YC-DREAL's picture
Model save
5ddf525 verified
{
"total_flos": 0.0,
"train_loss": 0.0027827628985492153,
"train_runtime": 18959.1276,
"train_samples": 7500,
"train_samples_per_second": 0.396,
"train_steps_per_second": 0.004
}