ekzhang/jax-js
A machine learning framework bringing JAX/NumPy-compatible array operations to the browser via WebGPU and WebAssembly compilation.
Collecting fresh signals — velocity needs a few days of history.
star history
jax-js is a browser-based machine learning library that implements JAX and NumPy APIs in pure JavaScript. It translates array operations into a compiler representation, then synthesizes kernels as WebAssembly for CPU execution and WebGPU for GPU acceleration. The project targets maximum portability, running client-side in any modern browser without external dependencies.