JAX beginner tutorial

par Dr Nestor Demeure (NERSC / Berkeley)

mercredi 24 avr. 2024, 14:00 → 17:00 Europe/Paris

Amphithéâtre PGG (U.Paris Cité - Bat. Condorcet - Level -1)

JAX is a high performance computing framework in Python from Google that has gained a lot of popularity over the past few years.

It uses the same syntax as the numpy array library, but the exact same code is compiled specifically for the backend you have at your disposal (CPU, GPU, TPU). It also supports auto differentiation which is a powerful way of obtaining gradients, very useful for scientific applications.

However coding with JAX requires a bit of training and some investment, which is what we would like to provide for you here. 

We have the pleasure to have Nestor Demeure, engineer at NERSC (Berkeley supercomputing center) and JAX specialist, visiting in late April. He will give us a 3h in-person introduction to JAX, starting at a beginner level, with live exercices.

To make sure the tutorial can be delivered in proper conditions, registration is mandatory as the number of seats will be limited. We kindly ask that you register below if you plan to attend in person. 

JAX exercices + solutions - Nestor Demeure

JAX tutorial slides - Nestor Demeure

Numpy vs Jax usage - Pierre Chanial