diff --git a/.buildkite/test_areas/entrypoints.yaml b/.buildkite/test_areas/entrypoints.yaml index 2db4c5ad5a..499abcc6c8 100644 --- a/.buildkite/test_areas/entrypoints.yaml +++ b/.buildkite/test_areas/entrypoints.yaml @@ -3,6 +3,7 @@ depends_on: - image-build steps: - label: Entrypoints Unit Tests + device: h200_35gb key: entrypoints-unit-tests timeout_in_minutes: 25 working_dir: "/vllm-workspace/tests" @@ -15,6 +16,7 @@ steps: - pytest -v -s entrypoints/weight_transfer - label: Entrypoints Integration (LLM) + device: h200_35gb key: entrypoints-integration-llm timeout_in_minutes: 60 working_dir: "/vllm-workspace/tests" @@ -114,6 +116,7 @@ steps: - image-build-amd - label: Entrypoints Integration (Responses API) + device: h200_35gb key: entrypoints-integration-responses-api timeout_in_minutes: 50 working_dir: "/vllm-workspace/tests" diff --git a/tests/entrypoints/openai/responses/test_parsable_context.py b/tests/entrypoints/openai/responses/test_parsable_context.py index 8ff3a1eeae..3a17da6127 100644 --- a/tests/entrypoints/openai/responses/test_parsable_context.py +++ b/tests/entrypoints/openai/responses/test_parsable_context.py @@ -68,7 +68,7 @@ async def client(server): async def test_basic(client: OpenAI, model_name: str): response = await client.responses.create( model=model_name, - input="What is 123 * 456?", + input="What is 123 * 456? Answer with only the number.", temperature=0.0, ) assert response is not None