Learn JAX

Functional ML — pure functions, vmap, pytrees. The JAX way of thinking.

Forward and reverse mode, custom derivatives, gradient checkpointing, per-example gradients. The autodiff toolbox.

vmap, scan, while_loop, fori_loop, cond, switch — JAX's primitives for batching and control flow.

Distributions, reparameterization, sampling techniques, MCMC, gradient estimators. Randomness as a controllable resource.

jit details, memory and device control, sharding APIs (Mesh, PartitionSpec, shard_map), mixed precision. The performance and parallelism toolbox.

Custom pytree nodes, host interop callbacks, quantization patterns, jaxpr inspection, advanced debugging. Beyond the curriculum.

Linear algebra, decompositions, FFT, convolutions, sparse arrays, polynomial fitting, numerical stability. Scientific computing in JAX.

Profiler hooks, debug primitives, fixing common tracer/concretization errors, AOT compilation. Practical-skill problems for production work.

Gradient transforms, optimizer chains, schedules, weight decay, EMA, masking. The production optimizer library for JAX.

Modules, layers from scratch, attention, transformer architectures, training loops, lifted transforms. Production model code in JAX (Linen API).

Flax's modern object-oriented API. Stateful modules, eager-mode workflow, split/merge state, lifted transforms, training, sharding. Standalone path through the recommended API for new projects.