Open Internet by MindsNet

Blaizzy/mlx-vlm

MLX-VLM is a Python package for running Vision Language Models (VLMs) on Mac devices using MLX, enabling inference and fine-tuning. It supports various models and offers features like chat UI, multi-image chat, and quantization. This tool makes it easy to leverage powerful VLMs locally. Best for: Developers and researchers working with Vision Language Models on Mac devices, especially those with Apple Silicon. Use cases: Running and fine-tuning VLMs like IDEFICS, LLaVA, and PaLIgemma on a MacBook; Building a local chat UI for VLMs using Gradio; Experimenting with multi-image chat support for models like Pixart; Optimizing VLM performance with activation quantization and TurboQuant KV Cache Highlights: Native support for Apple Silicon, enabling efficient performance on Mac devices; Extensive model support, including IDEFICS, LLaVA, PaLIgemma, and more; User-friendly features like chat UI, multi-image chat, and vision feature caching

Computing & Technology, Computer Science, Artificial Intelligence