-
Notifications
You must be signed in to change notification settings - Fork 285
Pass Scheduler Config for VLM Pipeline in WhoWhatBenchmark. #2318
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.
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
Pass Scheduler Config for VLM Pipeline in WhoWhatBenchmark. #2318
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Pull Request Overview
This PR enhances the OpenVINO GenAI visual-text pipeline by allowing a continuous-batching configuration (cb_config
) to be passed and applied via the new scheduler API. Function signatures for load_visual_text_genai_pipeline
, load_visual_text_model
, and the top-level load_model
dispatcher were extended to accept and forward arbitrary kwargs.
- Add
cb_config
handling and createSchedulerConfig
when provided - Update pipeline instantiation to use
scheduler_config
and setATTENTION_BACKEND
- Propagate
**kwargs
throughload_visual_text_model
andload_model
Comments suppressed due to low confidence (3)
tools/who_what_benchmark/whowhatbench/model_loaders.py:219
- This function now accepts
cb_config
withinkwargs
, but there’s no docstring describing it. Please add or update the docstring to explain thecb_config
parameter, its expected structure, and how it controls continuous batching.
def load_visual_text_genai_pipeline(model_dir, device="CPU", ov_config=None, **kwargs):
tools/who_what_benchmark/whowhatbench/model_loaders.py:226
- [nitpick] The key
cb_config
is a bit ambiguous—consider renaming it toscheduler_config
orbatching_config
for clarity and consistency withget_scheduler_config_genai
.
is_continuous_batching = kwargs.get("cb_config", None) is not None
tools/who_what_benchmark/whowhatbench/model_loaders.py:226
- The new continuous-batching code path isn’t covered by existing tests. Please add unit tests for when
cb_config
is provided and verify thatscheduler_config
is applied correctly.
is_continuous_batching = kwargs.get("cb_config", None) is not None
try: | ||
import openvino_genai | ||
except ImportError as e: | ||
logger.error("Failed to import openvino_genai package. Please install it. Details:\n", e) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Passing the exception as a separate argument may not include stack trace in all logging setups. Consider using logger.exception(...)
or logger.error(..., exc_info=True)
to ensure the error details are captured.
logger.error("Failed to import openvino_genai package. Please install it. Details:\n", e) | |
logger.error("Failed to import openvino_genai package. Please install it. Details:\n", exc_info=True) |
Copilot uses AI. Check for mistakes.
7d14750
No description provided.