Apple's machine learning team releases MLX, a framework for training and deploying machine learning models on Apple Silicon, on GitHub



While major technology companies such as Google, Meta, and Microsoft are actively developing AI, Apple is conservative about AI and is seen as lagging behind in the AI development race. In December 2023, Apple released the machine learning framework `` MLX '' for Apple Silicon , its own processor, on the software development platform GitHub.

GitHub - ml-explore/mlx: MLX: An array framework for Apple silicon
https://github.com/ml-explore/mlx



Apple launches MLX machine-learning framework for Apple Silicon | Computerworld
https://www.computerworld.com/article/3711408/apple-launches-mlx-machine-learning-framework-for-apple-silicon.html

Apple drops new MLX machine learning framework for Apple silicon Macs - 9to5Mac
https://9to5mac.com/2023/12/06/mlx-machine-learning-apple-silicon-mac/

Apple joins AI fray with release of model framework - The Verge
https://www.theverge.com/2023/12/6/23990678/apple-foundation-models-generative-ai-mlx

Many people thought that ``Apple was passive about AI,'' but in July 2023, it was reported that Apple was using its own chat AI ``Apple GPT'' in internal operations. In addition, in November, CEO Tim Cook said, ``Apple is researching generative AI. I can't explain in detail what that is, but I think we're investing in this area.'' There is no doubt that there are,' he said.

Apple CEO Tim Cook once again says, ``We are working responsibly to develop generative AI'' - GIGAZINE



Then, in December, Apple suddenly announced the machine learning framework XLM and released the source code on GitHub. Awni Hannun, a machine learning researcher at Apple, wrote in a post on X (formerly Twitter): 'Just in time for the holiday season, we are releasing new software from Apple's machine learning research team. 'It's an efficient machine learning framework designed specifically for (i.e. laptops).'



According to the MLX documentation , MLX is a NumPy- like framework designed for efficient and flexible machine learning on Apple silicon, and that the Python API closely follows NumPy with a few exceptions. About.

Apple says, 'MLX was designed by machine learning researchers, for machine learning researchers. The framework is intended to be user-friendly while efficiently training and deploying models. The design of the framework itself is also conceptually simple. We aim to make it easy for researchers to extend and improve MLX and explore new ideas quickly.'

Frameworks such as PyTorch , Jax , and ArrayFire have influenced the design of MLX, but the major difference between these and MLX is that MLX is a 'model for unified memory ' of Apple silicon. Since the MLX array exists in shared memory, it can be executed on any supported device (CPU and GPU at the time of article creation) without having to copy data each time it is operated.

Hannun also explained that by using MLX, it is possible to run Llama, train the Transformer language model, fine-tune it with LoRA , generate text with Mistral , generate images with Stable Diffusion, and recognize speech with Whisper . Did.



Related Posts:

in Software, Posted by log1h_ik