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

9 lines
239 B
JSON
Raw Permalink Normal View History

{
"epoch": 5.0,
"eval_accuracy": 0.9471856009625939,
"eval_loss": 0.16564106941223145,
"eval_runtime": 0.5723,
"eval_samples_per_second": 8.737,
"eval_steps_per_second": 3.495,
"num_input_tokens_seen": 2242920
}