Glossary · Deep learning
What is JAX? Definition and meaning
A library for array computing with automatic differentiation, JIT compilation and easy parallelism on accelerators.
Definition
JAX: A library for array computing with automatic differentiation, JIT compilation and easy parallelism on accelerators.
In short: Composable numerical computing with autodiff.
When it is used
Research and TPU-scale training.
Where it fits
JAX belongs to Frameworks in the Deep learning topic. These are the other terms in the same group:
- PyTorch: A flexible framework with dynamic computation graphs and strong GPU support.
- PyTorch Lightning: A library that organizes training loops, distributed training and logging.
- TensorFlow: An end-to-end platform with a graph runtime and deployment tooling.
- Keras: A simple API for defining and training neural networks across several backends.
- Hugging Face: A platform and set of libraries for models, datasets and training tools such as Transformers.
Learn more
The full guide to JAX explains how it works, when to use it and how it compares with the alternatives, with examples. The glossary lists every term on the site from A to Z.