Lie detector compiled into Qwen3.5 weights flags 91% of lies with zero false positives
teortaxesTex · x · 2026-10-29
A developer showcases Qwenputer: a full RISC-V CPU embedded in Qwen3.5-4B's weights, using Gated DeltaNet layers as RAM to run C and MicroPython inside forward passes (DeepMind math: 97.5–100%). A lie detector and its log are also compiled into the weights where the model can't tamper with them — flagging 91% of lies before the false value, with zero false positives across 28,073 tokens, and saying "the monitor is off" changes nothing.
More from Models
- DeepSeek clusters apart from GPTs on Artificial Analysis personality map — teortaxesTex · 2026-10-29
- ChatGPT users report Computer history broken for weeks despite being on — RileyRalmuto · 2026-10-29
- You.com: agentic search failures are behavior problems, fixed in Nemotron 3.5 Lightning — RichardSocher · 2026-10-29
- How I AI agent bench: Grok Bot wins big at 93%, Claude Duet trails at 79% — lennysan · 2026-10-29
- DeepMind slammed for announcing models weeks before availability amid Gemini Xenon chatter — adonis_singh · 2026-10-29
- Alphabet Q3 call: Gemini Carbon frontier model, MCP apps in Search, AI Mode ads expand — lilyraynyc · 2026-10-29