跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.17763 · v1

Knowledge After Collapse: Context and Wording Shape Factual Accessibility in Language Models

首次提交 2026-11-13 13:19
版本更新 2026-11-13 13:19
雷达观测 2026-11-17 04:57
arXiv ↗PDF ↗

摘要

When a language model accepts a false correction in a factual discussion, its next answer can depend on how the conversation continues. We introduce Knowledge Answer Divergence (KAD), an event-conditioned evaluation protocol that distinguishes initial false-answer adoption from subsequent persistence, recovery, and abstention. KAD freezes the conversation at the first explicit adoption of a designated false target and branches independent probes from that exact history. This enables matched comparisons of post-adoption behavior and history-management interventions that a single susceptibility score cannot capture. We evaluate eight local checkpoint conditions and six hosted endpoints, alongside an audited 300-question rerun. In corrected Llama 3.2 Instruct 3B controls, a neutral-history branch yields 42/46 correct responses, compared with 16/46 under full history. A matched cue-placement control changes false adoption by 27.3 percentage points in Qwen 3.5 4B. Twelve independently authored pressure templates demonstrate substantial sensitivity to wording. Together, these findings demonstrate the utility of KAD for evaluating factual reliability beyond initial susceptibility, enabling systematic comparisons of false-answer persistence, recovery conditions and conversational interventions.

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

目前只收录了一个版本。收录下一版后,可以对照标题、作者、分类和摘要。

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。

v1 · 2026-11-13 13:19 · 最新观测版本

Knowledge After Collapse: Context and Wording Shape Factual Accessibility in Language Models

When a language model accepts a false correction in a factual discussion, its next answer can depend on how the conversation continues. We introduce Knowledge Answer Divergence (KAD), an event-conditioned evaluation protocol that distinguishes initial false-answer adoption from subsequent persistence, recovery, and abstention. KAD freezes the conversation at the first explicit adoption of a designated false target and branches independent probes from that exact history. This enables matched comparisons of post-adoption behavior and history-management interventions that a single susceptibility score cannot capture. We evaluate eight local checkpoint conditions and six hosted endpoints, alongside an audited 300-question rerun. In corrected Llama 3.2 Instruct 3B controls, a neutral-history branch yields 42/46 correct responses, compared with 16/46 under full history. A matched cue-placement control changes false adoption by 27.3 percentage points in Qwen 3.5 4B. Twelve independently authored pressure templates demonstrate substantial sensitivity to wording. Together, these findings demonstrate the utility of KAD for evaluating factual reliability beyond initial susceptibility, enabling systematic comparisons of false-answer persistence, recovery conditions and conversational interventions.

查看此版本解读 →查看此版本原文 ↗