Handwired compiles plain JavaScript into an exact, fully differentiable neural net
ctjlewis · x · 2026-10-16
Chenglou released Handwired, a toy project that compiles ordinary JavaScript code into a neural network: no training required, outputs exactly match the original JS program, fully differentiable, and runnable in the browser without a GPU.
Discussion in the thread notes the theoretical grounding: functions form a closed cartesian category (simply-typed lambda calculus), and neural networks form one too, so the two can be compiled into each other — Siemens previously did similar work with the concat plugin compiling Haskell to ML. Quip: "the JavaScript crowd has discovered compiling to categories."
Related event: Handwired Compiles Plain JavaScript into a Neural Network Without Training(2 posts)→
More from coding & agent
- Developer rebuilds GitHub in 4 days: Keel makes agents declare intent before changing code — threepointone · 2026-10-16
- Stop using small models as judges with overloaded context, dev warns — xeophon · 2026-10-16
- Meta's open-weight Muse Spark Open hits Ollama with 1M-token context and Claude Code support — ollama · 2026-10-16
- Monid launches: one balance, pay-per-call access to 1,700+ tools for AI agents — HeyAmit_ · 2026-10-16
- OpenAI publishes Sites tutorial: build and share your first Site — OpenAIDevs · 2026-10-16
- OpenAI Sites goes GA for Enterprise with US/EU data residency — OpenAIDevs · 2026-10-16