A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 12:15 UTC
[CI][TPU] Skip structured outputs+spec decode tests on TPU (#17510)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
Apr 30, 2025 at 21:36 UTC
17b4d85f63e0edcf55cc7abc0769c3d0a9cbe340
1 file changed
+3
tests/v1/entrypoints/llm/test_struct_output_generate.py
+3
@@ -81,6 +81,9 @@ def test_structured_output(
81
):
82
monkeypatch.setenv("VLLM_USE_V1", "1")
83
84
+ if current_platform.is_tpu() and speculative_config:
85
+ pytest.skip("TPU does not support speculative decoding")
86
+
87
# Don't use eager execution on TPUs because we want to test for no
88
# recompilation at runtime
89
enforce_eager = bool(not current_platform.is_tpu())