Open
Conversation
mzr1996
requested changes
Mar 17, 2026
Comment on lines
+847
to
+848
| "llava_v1.5_7b_hf": partial(LLaVA, model_path="llava-hf/llava-1.5-7b-hf"), | ||
| "llava_v1.5_13b_hf": partial(LLaVA, model_path="llava-hf/llava-1.5-13b-hf"), |
Collaborator
There was a problem hiding this comment.
Why is the LLaVA_HF class not used in the config.py?
Author
There was a problem hiding this comment.
Thank you for pointing out the typo. It should indeed be LLaVA_HF here.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR introduces a new llava_hf class to support the native Hugging Face implementations of LLaVA (e.g., llava-hf/llava-1.5-7b-hf or llava-hf/llava-1.5-13b-hf).
Currently, loading an official llava-hf model using the default llava class results in a cascade of errors (missing image_mean attributes, device-side asserts, and pixel_values argument rejections). This is because the Hugging Face implementation handles multimodal inputs fundamentally differently than the original custom Haotian Liu architecture.
To prevent others from suffering through dependency conflicts and tensor mismatches, I've isolated the HF-specific logic into its own class.