1. What is JAX ? https://jax.readthedocs.io/en/latest/
JAX: High-Performance Array Computing — JAX documentation
Transformations JAX includes composable function transformations for compilation, batching, automatic differentiation, and parallelization.
jax.readthedocs.io
JAX is a python library for accelerator-oriented array computation and program transformation, designed for high-performance numerical computing and large-scale machine learning.
If you're looking to train neural networks, use Flax and start with its documentation. Some associated tools are Optax and Orbax. For an end-to-end transformer library built on JAX, see MaxText.
2. How JAX? https://jax.readthedocs.io/en/latest/notebooks/quickstart.html
JAX Quickstart — JAX documentation
JAX Quickstart JAX is NumPy on the CPU, GPU, and TPU, with great automatic differentiation for high-performance machine learning research. With its updated version of Autograd, JAX can automatically differentiate native Python and NumPy code. It can differ
jax.readthedocs.io
JAX is NumPy on the CPU, GPU, and TPU, with great automatic differentiation for high-performance machine learning research.