Apple has released the MLX machine learning framework for Apple Silicon

By: Bohdan Kaminskyi | 08.12.2023, 17:29

Zhiyue/Unsplash.

Apple has unveiled frameworks and model libraries designed to run on its chips.

Here's What We Know

Apple's machine learning research group has released MLX, a framework for running models efficiently on Apple Silicon chips. The company also introduced MLX Data library with pre-trained deep learning models.

Both products are available in open access on GitHub and PyPI resources.

According to the developers, when creating MLX, they focused on such popular frameworks as PyTorch, Jax and ArrayFire. Unlike them, however, MLX performs all tasks on supported devices (currently CPUs and GPUs) without moving data.

According to Computerworld, the framework is easy to learn but powerful enough to train complex AI models like Meta's Llama and Stable Diffusion.

According to Apple's Awni Hannun, MLX Data is a framework-independent package for efficient data loading. It is compatible with MLX, PyTorch or Jax.

The frameworks presented are focused on machine learning rather than popular generative AI applications. That said, the tech giant tries to avoid the very word "AI" in its keynote presentations.

Source: The Verge