| 1 | # For hf script, without -t option (tensor parallel size). |
| 2 | # bash .buildkite/lm-eval-harness/run-lm-eval-mmlupro-vllm-baseline.sh -m meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 -l 250 -t 8 -f 5 |
| 3 | model_name: "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8" |
| 4 | rocm_safetensors_load_strategy: prefetch |
| 5 | required_gpu_arch: |
| 6 | - gfx942 |
| 7 | - gfx950 |
| 8 | tasks: |
| 9 | - name: "mmlu_pro" |
| 10 | metrics: |
| 11 | - name: "exact_match,custom-extract" |
| 12 | value: 0.80 |
| 13 | limit: 250 # will run on 250 * 14 subjects = 3500 samples |
| 14 | num_fewshot: 5 |
| 15 | rtol: 0.05 |