大模型当裁判时，信任评分与事实判断真能独立吗？这项研究用受控QA测试揭开了两者的纠缠关系
arXiv:2608.21097v1 Announce Type: new Abstract: LLM-as-Judge systems can produce multi-dimensional evaluations, such as trustworthiness, reliability, …
研究大模型在事实性主张上的分歧，帮你看清不同LLM的可靠性差异与潜在盲点。
Article URL: https://zenodo.org/records/21829261 Comments URL: https://news.ycombinator.com/item?id=49272730 Points: 1 # Comments: 1
探究推理链长度如何影响大模型对答案事实性的判断，揭示关键机制。
arXiv:2604.06756v2 Announce Type: replace Abstract: Large language models (LLMs) has been widely adopted as a scalable surrogate for human evaluation,…
大模型如何在激活空间内编码“上下文真值”？这项研究揭示了线性方向的全新机制，带你深入AI判断真相的内部逻辑。
arXiv:2608.03035v1 Announce Type: new Abstract: Prior work has shown that LLMs encode the truth of factual propositions along linear directions in act…
3D脑MRI报告生成事实性难题，用成对图文检索与文本增强巧妙破解，医学影像AI必读。
arXiv:2411.15490v2 Announce Type: replace-cross Abstract: Acute ischemic stroke (AIS) requires time-critical decision-making, where inaccurate interpr…
揭秘大模型“记忆检索”瓶颈：参数化事实性缺失，究竟是模型没存住还是没找到？
arXiv:2602.14080v2 Announce Type: replace-cross Abstract: Standard factuality evaluations of LLMs treat all errors alike, obscuring whether failures a…
用科学数据微调大模型反而更易“信口开河”？最新多领域评估揭示幻觉风险，警示科学应用需谨慎。
arXiv:2606.21359v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used to communicate and explain scientific concepts, yet…
针对MCP-based LLM智能体的来源感知事实性验证新方法，专治AI幻觉，提升输出可信度。
arXiv:2606.18037v1 Announce Type: new Abstract: Tool-using LLM agents increasingly use the Model Context Protocol (MCP) to answer from heterogeneous e…
将人类应试技巧引入模型判断声明事实性，有效提升大模型在证据支撑下的核查准确率。
arXiv:2605.29712v1 Announce Type: cross Abstract: Grounded claim factuality checking is important for large language model (LLM) applications such as …
研究揭示LLM在训练数据中植入错误信念后，即使明确警告也无法纠正，警示AI安全与事实性漏洞。
Fine-tuning tests show "bias ... toward confidently representing the claims as true."
精准揭示LLM幻觉根源，区分两种违规类型，为模型知识归因与纠错提供新思路。
arXiv:2602.22787v2 Announce Type: replace-cross Abstract: Large language model (LLM) hallucinations, meaning fluent but factually incorrect generation…
综述RAG系统可信度挑战，涵盖事实性、鲁棒性与公平性等关键维度。
arXiv:2409.10102v2 Announce Type: replace-cross Abstract: Retrieval-Augmented Generation (RAG) has quickly grown into a pivotal paradigm in the develo…