LLM writes 355-weight 4-digit multiplier without training — 40,000x smaller than SOTA
lateinteraction · x · 2026-11-07
Tula's Braden Hancock spotlights the new "weight space programming" paradigm kicked off by lateinteraction's lab blog: transformers struggle to learn 4-digit multiplication, but an LLM wrote a solution in just 355 weights (40,000x fewer than SOTA) with 100% accuracy, without any training.
Hancock notes matrix-multiplier-based long multiplication isn't surprising per se, but with billions invested in matrix-multiply-optimized hardware, converting software 1.0 (traditional code) into software 2.0 (weights as code, now directly authored by superior models rather than trained) could unlock 40,000x-class speedups elsewhere.
More from coding & agent
- CodeRabbit podcast: ship less, not more, in the AI era — nutlope · 2026-11-07
- 16 MCP servers, 35 parallel agents — and it still ships slop — weswinder · 2026-11-07
- Creator apps like Instagram Edits and Roblox Studio are shipping with built-in MCPs for agents — trq212 · 2026-11-07
- Cartesian agent builds a 289-part Dymaxion car 3D model live, researching the design online — carlosbannon · 2026-11-07
- ArkenOS visualizes hundreds of autonomous agents collaborating as a 3D research swarm — _AustinCalvert_ · 2026-11-07
- AI agents blocked by Zillow, Indeed, Reddit — a scraping API got 3/3 through in testing — eyishazyer · 2026-11-07