Keras 3.0 deep learning API backs TensorFlow, PyTorch, Jax
Keras 3.0, a “full rewrite” of the Keras deep learning API, has arrived, providing a new multi back-end implementation of the API.
Unveiled November 27, and accessible from GitHub, Keras 3.0 enables developers to run Keras workflows on top of the Jax, TensorFlow, or PyTorch machine learning frameworks, featuring large-scale model training and deployment capabilities. Keras is deployed as a low-level cross-framework language to develop custom components such as layers, models, or metrics that can be used in native workflows in all three frameworks, with one codebase.