JAX: When NumPy Isn't Enough

Name: JAX: When NumPy Isn't Enough
Start: 2025-03-11T15:30:00-04:00
End: 2025-03-11T17:00:00-04:00
Location: Private Location (sign in to display)

by

Training/Workshop Programming Languages Research & Data Analysis

Tue, Mar 11, 2025

3:30 PM – 5 PM EDT (GMT-4)

Private Location (sign in to display)

31

Registered

Registration

Registration is now closed (this event already took place).

Details

GitHub repo: https://github.com/pfackeldey/Princeton-JAX-Workshop-2025-03-11

JAX is a Python library for high-performance array computations. It adheres to the Python Array API and is therefore a drop-in replacement for NumPy. It can outperform NumPy in many ways: it is accelerator-oriented, supports JIT compilation and automatic differentiation. These advantages make JAX a rising star in the world of deep learning. However, these benefits also come at the price of some flexibility.

This workshop introduces the JAX project, shows when it makes sense to use JAX instead of NumPy and discusses the sharp bits of JAX.

Format: Presentation and hands-on with Google Colab

Speakers

Peter Fackeldey

Hosted By

Research Computing | View More Events
Co-hosted with: GradFUTURES, Research Computing (OWNER)