Loop transformer hits mid-training bifurcation as fractal basins emerge alongside generalization
Stefania_druga · x · 2026-09-10
Researcher wgilpin0 shows that training a loop transformer on integer linear regression produces a mid-training bifurcation: the model figures out how to solve unseen examples at the same moment fractal basins appear in its loss landscape. The visualization offers a fresh lens on when and how transformers generalize.
More from Research
- Token Collapse: Study Finds Graph Soft Prompting Tokens in LLMs Fail to Encode Structure — Sapienza · 2026-10-27
- Mathathon 2.0 launches with Terence Tao's SAIR, $1.5M in compute for teams — _sathvikr · 2026-10-27
- The 'sweet lesson' is real: two kinds of ML problems need opposite strategies — mike64_t · 2026-10-27
- PixGR: Pixel-space generative decoding for fast feed-forward 3D reconstruction — kwangmoo_yi · 2026-10-27
- UniAction: unified action conditioning lets one world model handle mixed action types — ZeYanjie · 2026-10-27
- Infini-AI Lab open-sources Hermit, a framework for agent harness evolution and continual learning — BeidiChen · 2026-10-27