当标准探测准确率饱和时，引入“脆弱性”度量作为互补指标，为LLM预训练分析提供新视角。
arXiv:2606.11375v1 Announce Type: cross Abstract: Standard linear probing declares a property "encoded" when a classifier on hidden states achieves hi…
大模型能编码因果方向，但输出却返回反常识答案，揭示隐藏状态与表达的不匹配。
arXiv:2605.25891v1 Announce Type: cross Abstract: We find a mismatch between what large language models encode about a causal question and what they a…
基于LLM隐藏状态探测的多语言幻觉检测新方法，有效应对非英语场景的可靠性挑战。
arXiv:2605.24919v1 Announce Type: new Abstract: Hallucinations in Large Language Models (LLMs) represent a critical barrier to their reliable deployme…
几何引导的隐藏状态替换，揭秘扩散模型在语言模型中的最佳插入位置，DiHAL创新方案提升性能。
arXiv:2605.14368v1 Announce Type: cross Abstract: Continuous diffusion language models lag behind autoregressive transformers, partly because diffusio…
提出用JEPAs审计LLM微调：预测隐含表示而非输出，以提升任务指标。
arXiv:2605.15394v1 Announce Type: cross Abstract: Joint-embedding predictive architectures (JEPAs) propose that a model should learn more useful abstr…