Multi-framework modeling with KerasCV and KerasNLP

In this talk we demonstrated how to use the Keras domain packages KerasCV and KerasNLP to build production-grade applied ML workflows. Both packages are based on the new Keras Core library, which automatically brings multi-backend support for TensorFlow, JAX, and PyTorch to all your workflows in KerasCV and KerasNLP. We'll show you how to combine components from your favorite backend-specific package with multi-backend Keras, as well as how to easily train your Keras models on TPUs with JAX.

backend build core domain framework jax keras keras core kerascv library modeling production pytorch show support talk tensorflow workflows