Cheng Lou's Handwired Turns JavaScript into a Neural Net, No Training Needed
Summary
Cheng Lou publishes Handwired, a tool that rewrites regular JavaScript into matrix multiplication plus ReLU so programs stay differentiable without training, and it runs on JS or WebGPU. The MIT-licensed repository has 1.4k stars, and a demo claims JSON search over 100,000 documents in milliseconds.
Key Points
- GitHub shows 102 forks on the repository.
- Cheng Lou claims unfreezing the weights and training them is one possible use case.
- Cheng Lou claims the JSON search demo gets GPU-level parallelism from single-threaded JavaScript code.