27B Model Rewrites Its Own Weights Without Training or Gradients, Paper Claims
udmrzn · x · 2026-11-18
A new arXiv paper by Jiajun Fan et al., Language Models Can Rewrite Their Own Weights, shows a 27B language model can diagnose its own errors and rewrite its weights — with no training or gradients — to improve on unseen tasks.
The authors argue this hints models could be permanently updated through inference alone, bypassing the usual context-stuffed memory or expensive fine-tuning loops. If reliable on a single GPU, it opens a new paradigm for continuous, localized model adaptation, especially for self-improving agents.
Related event: Language Models Can Rewrite Their Own Weights Without Training(3 posts)→
More from Research
- MIT's Markus Buehler to receive 2027 SES medal for pioneering physics-aware AI — ProfBuehlerMIT · 2026-11-18
- Field experiment: Google AI Overviews cut publisher clicks 39.8% with no UX gain — paulnovosad · 2026-11-18
- RoboDojo sim "not even wrong": blog exposes physics flaws in popular robot benchmark — ZeYanjie · 2026-11-18
- CoTAM: Machine-Oriented Image Compression That Preserves Multimodal AI Reasoning Performance — burkov · 2026-11-18
- Formal verification meets agentic AI: neurosymbolic approach pitched as a path to trustworthy agents — luislamb · 2026-11-18
- Claude correctly folds ubiquitin from a plain prompt, no specialized tools needed — iScienceLuvr · 2026-11-18