xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-04-05 12:37:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

add gemma3 to test

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon 2025-08-04 12:56:13 -07:00
parent 06fba5410c
commit a772948c9d
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tests/compile/piecewise/test_full_cudagraph.py
View File

@ -71,7 +71,8 @@ def llm_pair(request):
[
# Model names for the llm_pair fixture
"deepseek-ai/DeepSeek-V2-Lite",
"Qwen/Qwen2-1.5B-Instruct"
"Qwen/Qwen2-1.5B-Instruct",
"google/gemma-3-1b-it",
],
indirect=True)
@pytest.mark.skipif(current_platform.get_device_capability() != (9, 0),
@ -126,6 +127,8 @@ class TestFullCUDAGraph:
("Qwen/Qwen2-1.5B-Instruct", True),
# MLA does not support capturing CUDA Graphs with size > max_num_seqs
("deepseek-ai/DeepSeek-V2-Lite", False),
# Full CUDA graph supports mixed full and sliding window attention.
("google/gemma-3-1b-it", True),
])
@pytest.mark.skipif(current_platform.get_device_capability() != (9, 0),
reason="Only Hopper GPUs support FA3 and FlashMLA")