fix meta-reference, test vllm

2025-12-18 17:08:40 +00:00 · 2024-12-16 23:45:15 -08:00 · 2024-12-16 23:45:15 -08:00 · a30aaaa2e5
commit a30aaaa2e5
parent b75e4eb6b9
3 changed files with 13 additions and 3 deletions
--- a/llama_stack/providers/tests/inference/fixtures.py
+++ b/llama_stack/providers/tests/inference/fixtures.py
@ -113,6 +113,7 @@ def inference_vllm_remote() -> ProviderFixture:
                provider_type="remote::vllm",
                config=VLLMInferenceAdapterConfig(
                    url=get_env_or_fail("VLLM_URL"),
+                    max_tokens=int(os.getenv("VLLM_MAX_TOKENS", 2048)),
                ).model_dump(),
            )
        ],