MLXMLX

MLX

解决什么问题What problem it solves

苹果芯片上的机器学习框架，由 Apple 研究团队开源。A machine learning framework for Apple silicon, open-sourced by Apple's research team.

这是什么What it is

官方描述：面向 Apple silicon 的机器学习数组框架。由 Apple 机器学习研究团队出品。

特点是统一内存模型：数组驻留共享内存，CPU 与 GPU 上运算无需搬运数据。

The official description: an array framework for machine learning on Apple silicon, from Apple machine learning research.

Its notable feature is the unified memory model: arrays live in shared memory, so CPU and GPU operations need no data transfer.

为什么需要知道Why it matters

用 Mac 跑模型时值得知道它存在。官方并未声称统一内存能让 Mac 跑大模型。Worth knowing if you run models on a Mac. Note the docs do not claim unified memory runs large models.

现在能做什么What you can do now

出处Sources