Files
Qwen2.5-MATH-1.5B-GRPO-Best/all_results.json

8 lines
202 B
JSON
Raw Permalink Normal View History

{
"total_flos": 0.0,
"train_loss": 1.1265189547477097e-09,
"train_runtime": 46427.9339,
"train_samples": 5580,
"train_samples_per_second": 0.961,
"train_steps_per_second": 0.06
}