Dave5

Using its hardware prowess is an obvious step for Apple, but company watchers have to be excited by the quiet moves they are making on the software front. They just released MLX, a new (machine learning) ML framework for Apple Silicon. This isn’t surprising — during the M3 MacBook launch, Apple was pretty explicit in saying that it was targeting the AI and ML community with its high-end machines. 

https://om.co/2023/12/06/apple-makes-a-quiet-ai-move/

According to Apple on GitHub, frameworks like PyTorch, Jax, and ArrayFire inspired the design of MLX, with the notable difference of having a shared memory, meaning any task run on MLX works on supported devices (right now, CPUs and GPUs) without moving data. Computerworld reported that MLX is intended to be easy to use for developers but has enough power to train AI models like Meta’s Llama and Stable Diffusion. Frameworks and model libraries help power many of the AI apps in the market now.

Awni Hannun, a machine learning researcher with Apple, tweeted that MLX Data is a “framework agnostic, efficient, and flexible package for data loading” and works with MLX, PyTorch, or Jax frameworks.

https://www.theverge.com/2023/12/6/23990678/apple-foundation-models-generative-ai-mlx