keras — Keras 3 is a multi-backend deep learning framework that allows users to build and train models using JAX, TensorFlow, PyTorch, or OpenVINO.

Keras 3 is a multi-backend deep learning framework that allows users to build and train models using JAX, TensorFlow, PyTorch, or OpenVINO.

About this project

Keras 3 is a high-level deep learning framework designed for flexibility and performance across multiple backends. It supports JAX, TensorFlow, and PyTorch for training and inference, with additional inference-only support for OpenVINO. Key capabilities include: - Multi-Backend Support: Users can switch between JAX, TensorFlow, and PyTorch to leverage specific framework advantages, such as JAX's performance or TensorFlow's production ecosystem. - Broad Application Range: Supports the development of models for computer vision, natural language processing, audio processing, timeseries forecasting, and recommender systems. - Interoperability: Keras models can be integrated into native PyTorch Modules or JAX-native model functions. It also supports datasets from various formats, including tf.data.Dataset and PyTorch DataLoaders. - Scalability: Designed to scale from local laptops to large GPU or TPU clusters. - Backwards Compatibility: Acts as a drop-in replacement for tf.keras when using the TensorFlow backend.