tfjs

Why we included this project

TensorFlow.js is the practical way to get machine learning running anywhere JavaScript runs, which for most teams means the browser and Node.js. You can train models from scratch with its Keras-style layers API, or convert an existing Python-trained TensorFlow model and run it client-side, which keeps user data on the device and avoids round-tripping to a server. The backend story is what makes it flexible: WebGL, WebGPU, and WebAssembly for the browser, plus native TensorFlow bindings under Node, so you can trade bundle size against speed depending on your target. One thing to watch in server use is that the Node bindings are synchronous, so heavy inference can block the event loop in a production service.