Fastvlm #495
Open
Fastvlm #495
Conversation
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
- Add `test_fastvlm` function to validate new ML model functionality. - Introduce `FastVLMMultiModalProjector` and `FastVLM` model classes to support vision-text multimodal operations. - Enhance `convert.py` by adding CoreML vision tower support and new CLI options (`--only-llm`, `--skip-vision`). - Refactor utilities to handle CoreML vision tower loading and add `force_download` option. - Cleanup unused and redundant test cases in `test_utils.py`. - Add configuration for `llava_qwen2` in prompt and model utilities.
There was a problem hiding this comment.
Pull Request Overview
This PR adds FastVLM support to the mlx-vlm project, implementing a new vision-language model with unique Core ML vision tower loading capabilities. FastVLM uses a sideband loading mechanism to overcome Hugging Face cache compatibility issues with Core ML model packages.
- Add FastVLM model implementation with Core ML vision tower support
- Implement sideband loading system for Core ML model packages to work around HF cache limitations
- Update utilities and conversion scripts to support new model architecture
Reviewed Changes
Copilot reviewed 10 out of 11 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
| mlx_vlm/utils.py | Add Core ML import, model type mapping, and vision tower loading logic |
| mlx_vlm/tests/test_utils.py | Comment out stale quantization tests |
| mlx_vlm/tests/test_models.py | Add comprehensive FastVLM model tests |
| mlx_vlm/prompt_utils.py | Add message format support for llava_qwen2 model type |
| mlx_vlm/models/fastvlm/ | New FastVLM model implementation with language model and Core ML integration |
| mlx_vlm/hf_tools/mlpackage_cache.py | New Core ML package caching and resolution utilities |
| mlx_vlm/convert.py | Add conversion support with Core ML file copying and new CLI options |
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
Co-authored-by: Copilot <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Add FastVLM (https://huggingface.co/papers/2412.13303) support.