FastVLM: Efficient vision encoding for vision language models(github.com)
335 points by nhod | 68 comments
Apple's GitHub repository with the official implementation of FastVLM, an efficient vision encoding method for vision language models presented at CVPR 2025.