Skip to content

Fix/vllm processor cache for text only model #359

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
hiyouga merged 3 commits into from
Jun 17, 2025
Merged

Fix/vllm processor cache for text only model #359

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5c2c10d
[utils] handle models that do not provide a processor by using a try/…
cyc00518 Jun 17, 2025
d37f6e0
[rollout] Adjusted the vLLM rollout code so disable_mm_preprocessor_c…
cyc00518 Jun 17, 2025
e27f9c9
Update tokenizer.py
hiyouga Jun 17, 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
3 changes: 2 additions & 1 deletion verl/workers/rollout/vllm_rollout_spmd.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,6 +83,8 @@ def __init__(self, model_path: str, config: RolloutConfig, tokenizer: PreTrained
engine_kwargs = {}
if config.limit_images:
engine_kwargs["limit_mm_per_prompt"] = {"image": config.limit_images}
engine_kwargs["disable_mm_preprocessor_cache"] = True
Comment thread
hiyouga marked this conversation as resolved.


self.inference_engine = LLM(
model=model_path,
Expand All @@ -99,7 +101,6 @@ def __init__(self, model_path: str, config: RolloutConfig, tokenizer: PreTrained
disable_log_stats=config.disable_log_stats,
enforce_eager=config.enforce_eager,
disable_custom_all_reduce=True,
disable_mm_preprocessor_cache=True,
enable_chunked_prefill=config.enable_chunked_prefill,
enable_sleep_mode=True,
**engine_kwargs,
Expand Down