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

8 lines
202 B
JSON
Raw Permalink Normal View History

{
"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
}