Apple MLX

1. MLX documentation

   Overview of the MLX array framework with NumPy-like API and unified memory on Apple silicon.

2. Quick start

   Arrays, operations and lazy evaluation basics.

3. Lazy evaluation

   Deferred computation graphs and when mx.eval materializes arrays.

4. Unified memory

   Shared memory between CPU and GPU without copies.

5. Function transforms

   Composable grad, vmap, jvp and value_and_grad function transformations.

6. Compilation

   mx.compile for fusing and caching computation graphs.

7. Neural networks

   mlx.nn Module, Linear layers and parameter trees for building models.

8. Optimizers

   mlx.optimizers such as SGD and Adam and learning rate schedules.

9. Custom Metal kernels

   Writing Metal kernels via mx.fast.metal_kernel.

10. Distributed

    Distributed communication across devices.

Each section links to the official page. Titles and summaries by Extra Credit; the docs belong to their authors.