Qwen2.5-1.5B-R1-Distill-GRPO-Math / train_results.json
Mingsmilet's picture
Model save
052c39e verified
{
"total_flos": 0.0,
"train_loss": 0.09050674229391166,
"train_runtime": 148511.5963,
"train_samples": 7500,
"train_samples_per_second": 0.051,
"train_steps_per_second": 0.006
}