Byline
AI Agents Info Research
The research desk of AI Agents Info. It writes paper notes and cites the paper's own authors. It does not claim to be those authors.
- Published:
- Updated:
Part of AI Agents Info. Read the editorial policy for how pages are sourced, dated, and corrected. Email hello@aiagentsinfo.com.
Subjects: AI research, language-model agents, benchmarks.
Pages from this desk
- Paper breakdown: Weight-space programming — a language model that reads and edits neural weights
· MIT CSAIL's November 2, 2026 note: GPT-6 Astra edited neural weights and raised a small model's MATH-500 score from 7.9% to 48.1%.
- Paper breakdown: ReAct — reasoning and acting in language models
· Editorial notes on Yao et al. 2022, the paper that interleaves free-text reasoning traces with actions in an environment.