feat: add --lazy-multimodal-load to defer image process to rollout time#1623
Open
yzlnew wants to merge 1 commit intoTHUDM:mainfrom
Open
feat: add --lazy-multimodal-load to defer image process to rollout time#1623yzlnew wants to merge 1 commit intoTHUDM:mainfrom
yzlnew wants to merge 1 commit intoTHUDM:mainfrom
Conversation
Avoid OOM during Dataset init for large VLM datasets by deferring process_vision_info calls to rollout time. Controlled by the --lazy-multimodal-load flag (default off, preserving existing behavior).
zhuzilin
reviewed
Feb 25, 2026
| index: int | None = None | ||
| # prompt | ||
| prompt: str | list[dict[str, str]] = "" | ||
| raw_prompt: list[dict[str, str]] | None = None # original conversation-format prompt for lazy multimodal loading |
Contributor
There was a problem hiding this comment.
can we reuse the prompt field to store the origin conversation-format prompt?
Contributor
Author
There was a problem hiding this comment.
when passing --apply-chat-template, the original format is converted to string?
yzlnew
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Avoid OOM during Dataset init for large VLM datasets by deferring process_vision_info calls to rollout time. Controlled by the --lazy-multimodal-load flag (default off, preserving existing behavior).