Files
Qwen2.5-0.5B-Math-SFT-1024/eval_results.json

7 lines
162 B
JSON
Raw Normal View History

{
"epoch": 3.0,
"eval_loss": 0.6683521866798401,
"eval_runtime": 0.3349,
"eval_samples_per_second": 131.393,
"eval_steps_per_second": 17.917
}