| 1 | group: LM Eval |
| 2 | depends_on: |
| 3 | - image-build-xpu |
| 4 | steps: |
| 5 | - label: "XPU GPQA Eval (GPT-OSS)" |
| 6 | key: xpu-gpqa-eval-gpt-oss |
| 7 | timeout_in_minutes: 60 |
| 8 | device: intel_gpu |
| 9 | agent_tags: |
| 10 | label: production |
| 11 | gpu: 1+ |
| 12 | mem: 24+ |
| 13 | no_plugin: true |
| 14 | working_dir: "." |
| 15 | env: |
| 16 | REGISTRY: "public.ecr.aws/q9t5s3a7" |
| 17 | REPO: "vllm-ci-test-repo" |
| 18 | VLLM_TEST_DEVICE: "xpu" |
| 19 | source_file_dependencies: |
| 20 | - csrc/ |
| 21 | - vllm/model_executor/layers/quantization |
| 22 | - tests/evals/gpt_oss/ |
| 23 | commands: |
| 24 | - >- |
| 25 | bash .buildkite/scripts/hardware_ci/run-intel-test.sh |
| 26 | 'cd tests && |
| 27 | pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-xpu.txt' |
| 28 | - label: LM Eval Large Models(2 GPUs) |
| 29 | key: lm-eval-large-models-2-gpus |
| 30 | timeout_in_minutes: 45 |
| 31 | device: intel_gpu |
| 32 | agent_tags: |
| 33 | label: production |
| 34 | gpu: 2+ |
| 35 | mem: 24+ |
| 36 | no_plugin: true |
| 37 | optional: true |
| 38 | working_dir: "." |
| 39 | env: |
| 40 | REGISTRY: "public.ecr.aws/q9t5s3a7" |
| 41 | REPO: "vllm-ci-test-repo" |
| 42 | VLLM_TEST_DEVICE: "xpu" |
| 43 | source_file_dependencies: |
| 44 | - vllm/ |
| 45 | - .buildkite/lm-eval-harness/ |
| 46 | commands: |
| 47 | - >- |
| 48 | bash .buildkite/scripts/hardware_ci/run-intel-test.sh |
| 49 | 'cd .buildkite/lm-eval-harness && |
| 50 | export VLLM_USE_DEEP_GEMM=0 && |
| 51 | pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-xpu-fp8.txt --tp-size=2' |