| 1 | model_name: "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16" |
| 2 | accuracy_threshold: 0.93 |
| 3 | num_questions: 1319 |
| 4 | num_fewshot: 5 |
| 5 | startup_max_wait_seconds: 1200 |
| 6 | env: |
| 7 | VLLM_ROCM_USE_AITER: "1" |
| 8 | server_args: >- |
| 9 | --enforce-eager |
| 10 | --max-model-len 4096 |
| 11 | --tensor-parallel-size 4 |
| 12 | --enable-expert-parallel |
| 13 | --mamba-backend triton |
| 14 | --moe-backend triton |
| 15 | --speculative-config '{"method":"mtp","num_speculative_tokens":5}' |