← 研踪 PaperTrailarXiv:2608.00828 · v2
Isotropy Cliffs: The Geometric Signature of Decision-Making in Large Language Models
首次提交 2026-08-02 03:11
版本更新 2026-11-11 23:45
雷达观测 2026-11-13 14:16
摘要
We investigate the geometry of decision-making in Multiple Choice Question Answering (MCQA) through the lens of isotropy. Analyzing five open-weight models across diverse datasets, we identify decision-critical transition layers characterized by a shift in isotropy, coinciding with a major representational change and the emergence of task-relevant clusters. We demonstrate that this synchronized geometric behavior is strongly correlated with downstream accuracy ($r\approx0.84$), displaying its relevance for successful decision-making. Furthermore, we show that this transition is robust to prompt variations, suggesting that it reflects a general mechanism of model behavior.
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-11 23:45 · 最新观测版本
Isotropy Cliffs: The Geometric Signature of Decision-Making in Large Language Models
We investigate the geometry of decision-making in Multiple Choice Question Answering (MCQA) through the lens of isotropy. Analyzing five open-weight models across diverse datasets, we identify decision-critical transition layers characterized by a shift in isotropy, coinciding with a major representational change and the emergence of task-relevant clusters. We demonstrate that this synchronized geometric behavior is strongly correlated with downstream accuracy ($r\approx0.84$), displaying its relevance for successful decision-making. Furthermore, we show that this transition is robust to prompt variations, suggesting that it reflects a general mechanism of model behavior.
查看此版本解读 →查看此版本原文 ↗