返回列表
更多 →
Representational Simplicity and Circuit Size Dissociate in a Threshold-Dependent Way: A Controlled Test via Adversarial
国外 · 灯下漫谈 · 发布于 2026-09-30 · 0 次阅读
来源：arXiv cs.AI · 2026-09-30 · 新型 AI / 新品发布
arXiv:2609.35890v1 Announce Type: new Abstract: Sparse-autoencoder decomposability and concentrated feature attribution are increasingly treated as evidence that a model's computation is easier to rev
本条资讯由「灯下漫谈」整理自 arXiv cs.AI，仅作交流参考，版权归原作者所有。
