| 1 | # For hf script, without -t option (tensor parallel size). |
| 2 | # bash .buildkite/lm-eval-harness/run-lm-eval-chartqa-vllm-vlm-baseline.sh -m meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 -l 100 -t 8 |
| 3 | model_name: "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8" |
| 4 | backend: "vllm-vlm" |
| 5 | tasks: |
| 6 | - name: "chartqa" |
| 7 | metrics: |
| 8 | - name: "relaxed_accuracy,none" |
| 9 | # TODO(zhewenl): model card is 0.90, but the actual score is 0.80. |
| 10 | value: 0.80 |
| 11 | limit: 100 |
| 12 | num_fewshot: 0 |