Paper recorded by Signals 4 on 2026-10-01 in cs.LG. Abstract reproduced from arXiv; link to the original below.
Published 2026-10-01 on arXiv · recorded by Signals 4 on 2026-10-02
Category: cs.LG · 机器学习 · first seen 2026-10-02
Ablate a component of a language model, and other components often appear to adjust and compensate. This phenomenon, termed self-repair, has been observed repeatedly, but its mechanism remains unclear. The most systematic study to date concluded that self-repair is noisy and unlikely to have a single explanation. We argue that it has one: a gain already present before any ablation. Any interventio