Apple has introduced MLX, a new framework specifically designed for the creation of machine learning models. This tool is tailored to run exclusively on Apple Silicon chips, ensuring that any applications derived from it remain unique to Apple’s ecosystem.
MLX is available for free on GitHub, but not on Apple’s official repository. Instead, it’s hosted on Awni Hannun’s page, an engineer from Apple’s Machine Learning Research team.
This framework is aimed at developers with a deep understanding of artificial intelligence, offering integrations with other popular tools in the field. For instance, it’s compatible with NumPy, a Python library crucial for creating multidimensional arrays for data organization, and also supports C++, providing flexibility in programming language choice.
One of the most exciting prospects of MLX is its potential to serve as a foundation for AI-driven generative applications, similar to ChatGPT. By utilizing Apple’s own hardware, MLX can operate independently of external processing resources. Inspired by other popular machine learning frameworks like PyTorch, Jax, and ArrayFire, MLX’s advantage lies in its ability to run on Apple Silicon chips, accessing their ultra-fast unified memory.
Apple’s cautious stance on AI, as articulated by CEO Tim Cook, is reflected in the development of MLX. The company is already employing MLX for its machine learning models and is reportedly developing its version of ChatGPT. This move is a part of Apple’s broader strategy to establish a presence in the AI field, an area where it has previously lagged behind competitors like Google, Microsoft, and Meta. These companies have been aggressively investing in AI and machine learning, leading to a surge in the popularity of generative AI tools.
The unconventional release of MLX, without an official announcement and through an employee’s personal GitHub repository, adds an element of intrigue to Apple’s foray into AI.