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 13, 17:00 UTC
[CI] Add compile/test_multimodal_compile.py to CI (#28151)
Signed-off-by: Yanan Cao <gmagogsfm@gmail.com> Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
gmagogsfm committed
Nov 5, 2025 at 21:41 UTC
bde5039325a0b60d7021d6945b5abad0d057c72a
2 files changed
+3
.buildkite/test-pipeline.yaml
+1
@@ -460,6 +460,7 @@ steps:
460
- tests/compile
461
commands:
462
- pytest -v -s compile/test_basic_correctness.py
463
+ - pytest -v -s compile/test_multimodal_compile.py
464
- pytest -v -s compile/piecewise/
465
466
- label: PyTorch Fullgraph Test # 22min
tests/compile/test_multimodal_compile.py
+2
@@ -4,10 +4,12 @@ import pytest
4
5
from vllm.compilation.counter import compilation_counter
6
from vllm.config.compilation import CompilationMode
7
+from vllm.platforms import current_platform
8
9
10
# forked needed to workaround https://github.com/vllm-project/vllm/issues/21073
11
@pytest.mark.forked
12
+@pytest.mark.skipif(not current_platform.is_cuda(), reason="Skip if not cuda")
13
def test_qwen2_5_vl_compilation(vllm_runner, monkeypatch):
14
"""Test that Qwen2.5-VL vision submodules are compiled.
15