Skip to content

[https://nvbugs/5542867][fix] Fix the non-determinism issue in the mm_encoder test #8033

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
merged 2 commits into from
Sep 29, 2025
Merged

[https://nvbugs/5542867][fix] Fix the non-determinism issue in the mm_encoder test #8033

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f197521
Fix batch size and lower the threshold to fight non-determinism in t…
chang-l Sep 26, 2025
6ef3945
Formatting
chang-l Sep 27, 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
11 changes: 6 additions & 5 deletions tests/unittest/_torch/multimodal/test_mm_encoder_standalone.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,7 +168,6 @@ def test_single_image_chat(model_key, multimodal_model_config):
"llava-v1.6-mistral-7b-hf",
])
def test_multi_request_batch_chat(model_key, multimodal_model_config):
pytest.skip("https://nvbugspro.nvidia.com/bug/5542867")
"""Test batching multiple multimodal requests and verify encoder path matches raw path.

This mirrors test_single_image_chat but with a batch of size 3.
Expand Down Expand Up @@ -200,10 +199,12 @@ def test_multi_request_batch_chat(model_key, multimodal_model_config):

encoder = MultimodalEncoder(model=encoder_model_dir,
max_batch_size=max_batch_size)
llm = LLM(model=encoder_model_dir,
backend='pytorch',
kv_cache_config=kv_cache_config,
trust_remote_code=True)
llm = LLM(
model=encoder_model_dir,
backend='pytorch',
kv_cache_config=kv_cache_config,
max_batch_size=1, # fix batch size to reduce non-determinism in tests
trust_remote_code=True)

config_path = os.path.join(llm._hf_model_dir, 'config.json')
assert os.path.exists(
Expand Down