Primary sourcearXiv cs.AI
View sourceEvent ·Multi-source
Theoria:基于非正式推理状态的改写接受性验证
Theoria 是一种验证架构,将候选解答重写为一系列类型化状态转换,每步转换需提供明确理由(引用、计算或给定事实),且可独立审计。在 HLE-Verified Gold(185 道文本专家题)上,Theoria 认证了 105 题,严格精度达 91.4%(Wilson 95% CI [84.5%, 95.4%])。与整体 LLM 评判者相比,两者错误分布差异大(Jaccard 0.14-0.36),可互补。在 95 个对抗性有毒证明上,结构评判者捕获 94.7%,高于整体评判的 83.2%(p=0.0017)。在 GPQA Diamond(n=65)上,认证精度为 97.1%。
Current read
Theoria 能审计 AI 回答的每一步推理,HLE 上精度 91.4%,比纯 LLM 评判更可追溯,专治隐藏前提和伪造引用。
3 sources63° AI heatLast updated 2026/7/1 17:56:42
Evidence
3 sourcesConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.