| 1 | model_name: "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4" |
| 2 | accuracy_threshold: 0.93 |
| 3 | num_questions: 1319 |
| 4 | num_fewshot: 5 |
| 5 | max_concurrency: 16 |
| 6 | rocm_request_timeout_seconds: 1800 |
| 7 | min_acceptance_length: 2.0 |
| 8 | startup_max_wait_seconds: 1200 |
| 9 | server_args: >- |
| 10 | --enforce-eager |
| 11 | --max-model-len 4096 |
| 12 | --max-num-seqs 16 |
| 13 | --max-num-batched-tokens 32768 |
| 14 | --gpu-memory-utilization 0.5 |
| 15 | --tensor-parallel-size 2 |
| 16 | --enable-expert-parallel |
| 17 | --mamba-backend triton |
| 18 | --speculative-config '{"method":"mtp","num_speculative_tokens":5}' |