Introduction to JAX for Deep Learning

Dive deeper into JAX's powerful functional transformations. This course covers explicit pseudo-random number generation for reproducibility, automatic vectorization with jax.vmap for batching, an introduction to jax.shard_map for multi-device parallelism, the concept of PyTrees for handling complex data structures, and basic profiling/debugging techniques.

This course guides you through building a simple Multi-Layer Perceptron (MLP) from scratch using only JAX and its NumPy API. You'll tackle the classic XOR problem, learning how to manage model parameters as PyTrees, implement the forward pass, define a loss function, compute gradients, and write a manual training loop with a basic optimizer step.

Elevate your JAX deep learning projects with Flax and Optax. This course introduces Flax for concisely defining neural network architectures as Modules, and Optax for a comprehensive suite of optimizers. You'll refactor the XOR classifier from the previous course, leveraging these powerful libraries for more structured and maintainable code.

Apply your JAX, Flax, and Optax skills to a practical project: building an image classification pipeline. This course covers setting up a multi-file project, loading and preprocessing image data (e.g., MNIST), defining a Convolutional Neural Network (CNN) with Flax, and implementing robust training and evaluation loops.