mlx-lm

Tech stack & purpose

MLX LM is a Python package for generating text and fine-tuning large language models on Apple silicon using the MLX framework. It integrates with Hugging Face Hub to provide easy access to thousands of pre-trained language models, and supports quantization, model uploading, low-rank fine-tuning with LoRA or QLoRA, and distributed inference. The package offers both command-line interfaces (like `mlx_lm.generate` and `mlx_lm.chat`) and a Python API for text generation, including streaming capabilities and advanced features like prompt caching and rotating key-value caches for handling long contexts.

Languages

Contributors (top 30 of 196)

• Awni Hannun291 commits
• Angelos Katharopoulos61 commits
• Gökdeniz Gülmez53 commits
• Anchen38 commits
• Tarjei Mandt32 commits
• Ivan Fioravanti25 commits
• Prince Canuma24 commits
• madroid17 commits
• Chime Ogbuji13 commits
• n8programs10 commits
• Neil Mehta9 commits
• Madroid Ma8 commits
• Alex Barron7 commits
• will-lms5 commits
• Pedro Cuenca5 commits
• Sugato Ray4 commits
• Shunta Saito4 commits
• christian-lms4 commits
• Yi Wang4 commits
• Josh York4 commits
• JosefAlbers3 commits
• Evan Quiney3 commits
• Thomas Lazarus3 commits
• Nathan Sashihara3 commits
• Matt Clayton3 commits
• Vincent Amato3 commits
• Daniel Nakov3 commits
• Ashish3 commits
• Chris McMaster3 commits
• 
  Ryan Goulden3 commits · 34 followers