Learn JAX with Real Code Examples

Compute gradient of a scalar function with `grad`

Train a simple neural network using JAX arrays and `grad`

Vectorize loss computation over a batch with `vmap`

JIT-compile a physics simulation for GPU execution

Parallelize reinforcement learning environment rollout across multiple GPUs using `pmap`

Ensure functions are pure (no side effects) for `jit`/`grad` compatibility

Check data types: JAX often requires float32 arrays for GPU

Debug uncompiled functions before applying `jit`

Use `jax.debug.print` for intermediate values

Update `jax` and `jaxlib` to compatible versions

Validate functions on small arrays before batching

Compare gradients with numerical approximations

Test JIT-compiled and vectorized functions separately

Ensure reproducibility with PRNG keys

Benchmark performance on target device

Run JAX computations on CPU/GPU/TPU

Export trained models parameters for Flax/Haiku

Integrate with production ML pipelines via XLA-compiled functions

Serve batch predictions using compiled functions

Use JAX in research simulations or cloud TPU workflows

NumPy for array operations

SciPy for scientific computation

Optax for gradient-based optimization

Flax/Haiku for neural network modeling

TensorFlow Datasets (TFDS) for dataset loading

Optax for optimizers

Flax/Haiku for high-level neural networks

TensorFlow and PyTorch interoperability via ONNX/XLA

GPU/TPU hardware for acceleration

NumPy and SciPy for scientific computation

Use JIT to accelerate heavy computations

Vectorize functions instead of Python loops

Keep functions pure for composability

Cache and reuse compiled functions

Leverage multi-device parallelism with `pmap`

Transitioning to functional programming mindset

Debugging JIT-compiled functions

Ensuring reproducibility with PRNG keys

Managing multi-device parallelism

Integrating JAX with larger ML frameworks for production