| 1 | # Qwen4Exp accuracy evaluation |
| 2 | |
| 3 | This suite starts a Qwen3.8-Flash-Next-FP8 OpenAI-compatible server once and |
| 4 | uses EvalScope to evaluate GSM8K and AIME25. |
| 5 | |
| 6 | ```bash |
| 7 | # B200 |
| 8 | pytest -s -v tests/evals/qwen4_exp/test_accuracy.py \ |
| 9 | --config-list-file=configs/models-b200.txt |
| 10 | |
| 11 | # H200 |
| 12 | pytest -s -v tests/evals/qwen4_exp/test_accuracy.py \ |
| 13 | --config-list-file=configs/models-h200.txt |
| 14 | ``` |