| 1 | model_name: "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8" |
| 2 | tasks: |
| 3 | - name: "gsm8k" |
| 4 | metrics: |
| 5 | - name: "exact_match,strict-match" |
| 6 | value: 0.7142 |
| 7 | - name: "exact_match,flexible-extract" |
| 8 | value: 0.4579 |
| 9 | moe_backend: "flashinfer_cutlass" |
| 10 | limit: 1319 |
| 11 | num_fewshot: 5 |
| 12 | max_model_len: 262144 |
| 13 | kv_cache_dtype: fp8 |
| 14 | enforce_eager: false |
| 15 | apply_chat_template: true |
| 16 | fewshot_as_multiturn: true |
| 17 | trust_remote_code: true |