diff --git a/tests/v1/entrypoints/llm/test_struct_output_generate.py b/tests/v1/entrypoints/llm/test_struct_output_generate.py index d2569959114..c3ea024f58c 100644 --- a/tests/v1/entrypoints/llm/test_struct_output_generate.py +++ b/tests/v1/entrypoints/llm/test_struct_output_generate.py @@ -81,6 +81,9 @@ def test_structured_output( ): monkeypatch.setenv("VLLM_USE_V1", "1") + if current_platform.is_tpu() and speculative_config: + pytest.skip("TPU does not support speculative decoding") + # Don't use eager execution on TPUs because we want to test for no # recompilation at runtime enforce_eager = bool(not current_platform.is_tpu())