Files
Qwen3-8B-SFT-envbench_gpt5-…/eval_results.json

9 lines
240 B
JSON
Raw Permalink Normal View History

{
"epoch": 5.0,
"eval_accuracy": 0.8568963869568719,
"eval_loss": 0.4850449860095978,
"eval_runtime": 4.2558,
"eval_samples_per_second": 11.749,
"eval_steps_per_second": 3.055,
"num_input_tokens_seen": 31362728
}