Observed arrival · 2026-10-17
Handwired: JavaScript as a learnable neural net
An open-source project guide with live demos showing how ordinary JavaScript can become an exact, differentiable neural net.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- JavaScript developers exploring differentiable programs
- Worth noticing
- The ranking demo compares four learnable formula weights with 65 generic-network numbers, using 200 seeded test searches.
Field notes
The guide describes Handwired programs as fixed lists of 32-bit float operations, using weighted sums, ReLU products, and exact helper operations. Its examples include fitting a simulated ball's gravity, drag, and bounce to a recorded path, plus a Mini OS described as having 9,909 weights across 130 layers. The ranking demo uses simulated shoppers and compares models after learning from clicks.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue