Skip to content

[CI] Tiny speed up VLM CI #14517

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Kangyan-Zhou merged 4 commits into from
Dec 7, 2025
Merged

[CI] Tiny speed up VLM CI #14517

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
0a87eeb
more
b8zhong Dec 5, 2025
96c6ee4
more
b8zhong Dec 5, 2025
69882db
more
b8zhong Dec 5, 2025
616d3d4
Merge branch 'main' into tiny-vlm-test-speedup
b8zhong Dec 6, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions python/sglang/test/mmmu_vlm_mixin.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,7 @@ def run_mmmu_eval(
model = "openai_compatible"
tp = 1
tasks = "mmmu_val"
batch_size = 32
batch_size = 64
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Dec 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The batch size is now hardcoded to 64 in two places: here as batch_size, and in _run_vlm_mmmu_test for the --cuda-graph-max-bs argument (line 140). This could lead to inconsistencies if one is updated but not the other. To improve maintainability, consider defining this value as a constant at the module level (e.g., VLM_BATCH_SIZE = 64) and reusing it in both locations. This would ensure they always stay in sync.

log_suffix = "openai_compatible"
os.makedirs(output_path, exist_ok=True)

Expand Down Expand Up @@ -137,7 +137,7 @@ def _run_vlm_mmmu_test(
other_args=[
"--trust-remote-code",
"--cuda-graph-max-bs",
"32",
"64",
"--enable-multimodal",
"--mem-fraction-static",
str(self.parsed_args.mem_fraction_static), # Use class variable
Expand Down
2 changes: 1 addition & 1 deletion test/srt/models/test_vlm_models.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@
MODELS = [SimpleNamespace(model="openbmb/MiniCPM-V-2_6", mmmu_accuracy=0.4)]
else:
MODELS = [
SimpleNamespace(model="google/gemma-3-27b-it", mmmu_accuracy=0.45),
SimpleNamespace(model="google/gemma-3-4b-it", mmmu_accuracy=0.38),
SimpleNamespace(model="Qwen/Qwen2.5-VL-3B-Instruct", mmmu_accuracy=0.4),
SimpleNamespace(model="openbmb/MiniCPM-V-2_6", mmmu_accuracy=0.4),
]
Expand Down
4 changes: 2 additions & 2 deletions test/srt/run_suite.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,7 @@
TestFile("models/test_qwen_models.py", 90),
TestFile("models/test_reward_models.py", 103),
TestFile("models/test_transformers_models.py", 245),
TestFile("models/test_vlm_models.py", 282),
TestFile("models/test_vlm_models.py", 270),
TestFile("openai_server/basic/test_openai_embedding.py", 70),
TestFile("openai_server/basic/test_openai_server.py", 184),
TestFile("openai_server/basic/test_protocol.py", 3),
Expand Down Expand Up @@ -237,7 +237,7 @@
TestFile("models/test_qwen_models.py", 82),
TestFile("models/test_reward_models.py", 132),
TestFile("models/test_transformers_models.py", 320),
TestFile("models/test_vlm_models.py", 437),
TestFile("models/test_vlm_models.py", 387),
TestFile("openai_server/basic/test_openai_embedding.py", 141),
TestFile("openai_server/basic/test_openai_server.py", 149),
TestFile("openai_server/basic/test_protocol.py", 10),
Expand Down
Loading