Development

A WebAssembly interpreter is hard-coded into transformer weights

development2026-03-16

Claim

Percepta hard-coded a WebAssembly interpreter into transformer weights, executing arbitrary C code as tokens for millions of steps, showing neural networks are practical general-purpose computers.

Reporting date: 2026-03-16 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to March 16, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to March 16, 2026

Actor

Percepta

Evidence for

Supersedes

A 121-parameter adder is hand-coded rather than trained

Referenced by

Cite or correct this record

Citation formatCitation text

A WebAssembly interpreter is hard-coded into transformer weights
Innermost Loop KB — secondary record
Reporting date: 2026-03-16
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-03-16-transformers-run-arbitrary-c-code
Claim: Percepta hard-coded a WebAssembly interpreter into transformer weights, executing arbitrary C code as tokens for millions of steps, showing neural networks are practical general-purpose computers.
Attributed sources:
- Welcome to March 16, 2026 — https://theinnermostloop.substack.com/p/welcome-to-march-16-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.