Looping soft tokens back into LLMs reveals emergent chaos, measurable via Lyapunov exponents
CatAstro_Piyush · x · 2026-10-23
tdj11100 reports a cross-disciplinary experiment: looping a language model's own soft tokens back into it produces an emergent landscape of order and chaos, and behaviors that superficially make no sense can be understood through chaos theory. The retweeted writeup by geogristle explains how to extract leading Lyapunov exponent sweeps from LLMs, giving a quantitative tool for this "AI as dynamical system" view.
More from Research
- Stanford and Databricks use AI agents to reproduce medical findings and test analytical robustness — ChengleiSi · 2026-10-23
- EMNLP paper: diffusion LMs internally encode latent denoising timestep representations — gsarti_ · 2026-10-23
- KU Leuven and VIB open two Group Leader professorships in AI for Biology, start 2027 — jpillowtime · 2026-10-23
- Subspace Tuning curbs LLM catastrophic forgetting by freezing principal singular directions — scottniekum · 2026-10-23
- Perceptic launches PercepticOS, an AI workspace for 'scientific superintelligence' — nathanbenaich · 2026-10-23
- EPFL's Bosselut to talk long context as test-time learning at ELLIS seminar — andre_t_martins · 2026-10-23