tfjs

TensorFlow.js is an open-source hardware-accelerated JavaScript library for training and deploying machine learning models. It provides flexible low-level APIs for numerical computation and neural networks, as well as a high-level layers API similar to Keras. The library supports multiple backends including CPU, WebGL, WASM, WebGPU, Node.js, and React Native, allowing models to run in browsers, Node.js, and mobile environments. Users can build models from scratch, import pre-trained TensorFlow SavedModel or Keras models using the converter, and retrain existing models with client-side data. The repository combines several packages: Core, Layers, Data, Converter, Vis, and AutoML. It also offers benchmark tools for measuring performance metrics locally or across remote devices. TensorFlow.js is part of the TensorFlow ecosystem and includes tutorials, examples, and a gallery of related projects.