MLX

MLX is an array framework for machine learning developed by Apple’s machine learning research team. This open-source framework is designed and optimized specifically for Apple Silicon chips, drawing inspiration from frameworks such as NumPy, PyTorch, Jax, and ArrayFire. It provides a simple and user-friendly approach to help developers effectively develop, train, and deploy models on Apple’s M-series chips.
The main functions of MLX
Familiar API: MLX has a Python API that closely follows NumPy. MLX also has a fully functional C++API, which is very similar to the Python API.
Composable function transformations: MLX supports composable function transformations for automatic differentiation, automatic vectorization, and computational graph optimization.
Lazy computation: The computation in MLX is lazy computation, and arrays are only concretized when needed.
Dynamic graph construction: Computational graphs in MLX are dynamically constructed. Changing the shape of function parameters does not trigger slow compilation, and debugging is simple and intuitive.
Multi device: can run on any supported device (CPU and GPU).
Unified Memory: The main difference between MLX and other frameworks is the unified memory model and array shared memory. The operations on MLX can run on any supported device type without the need to move data.