Anthropic用J-lens工具探入Claude Opus 4.6内部，发现了一个名为J-space的隐藏空间，揭示大模型处理概念的奇妙机制。
The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at what’s really going on inside large language models as t…
利用激活几何进行无监督特征挖掘，无需人类标注即可揭示大模型内部表征
arXiv:2607.04222v1 Announce Type: new Abstract: Interpretability methods aim to reveal the features represented inside large language models (LLMs). M…
开源模型能否解释闭源模型？这篇论文探讨替身保真度的条件与边界。
arXiv:2606.32008v1 Announce Type: new Abstract: Mechanistic interpretability (MI) requires full access to model internals, yet the APIs for most widel…
用LLM Agent自动发现并优化概念编辑算法，为模型可解释性开辟新路径
Article URL: https://dmodel.ai/concept-erasure/ Comments URL: https://news.ycombinator.com/item?id=48746983 Points: 3 # Comments: 0
对话LLM黑盒取证新方法，揭秘模型内部行为无需访问模型参数。
arXiv:2606.22698v1 Announce Type: cross Abstract: As LLM-powered scams proliferate, black-box forensics for conversational LLM agents offers a path to…
揭秘LLM电路发现中的方差来源，为可解释性研究提供新视角
arXiv:2606.16920v1 Announce Type: cross Abstract: Circuit discovery is a key technique in mechanistic interpretability to pinpoint the model component…
新论文提出多粒度节点剪枝方法，高效发现因果电路，提升模型可解释性。
arXiv:2512.10903v2 Announce Type: replace Abstract: Circuit discovery aims to identify minimal subnetworks that are responsible for specific behaviors…
提出 Jacobian Scopes 方法，实现大语言模型 token 级因果归因，论文已更新至 v3。
arXiv:2601.16407v3 Announce Type: replace-cross Abstract: Large language models (LLMs) make next-token predictions based on clues present in their con…
多模态大模型能听会看，但音频和视觉信号如何内部流动最终影响回答？这篇论文深入追踪其信息路径。
arXiv:2606.10147v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) can listen and see, but how do audio and visual signals actua…
新方法实现又快又忠实的函数向量，提升大模型行为控制的可信度与效率
arXiv:2606.05079v1 Announce Type: new Abstract: Function vectors (FVs) are task representations elicited during in-context learning that can be used t…
一项被KDD 2026接收的新方法，通过形式化与校准探针精准映射大模型推理能力，揭示LLM深层认知图谱。
arXiv:2603.05290v2 Announce Type: replace Abstract: Large language models (LLMs) achieve promising performance, yet their ability to reason remains po…
利用结构信息让大模型精准定位自身推理错误，提升可靠性与可解释性。
arXiv:2602.02416v2 Announce Type: replace Abstract: Self-correction in language models remains elusive. In this work, we explore whether language mode…
从结构视角解析大模型的多语言能力，揭示语言间共性与差异的深层机制。
arXiv:2606.01800v1 Announce Type: cross Abstract: Large language models (LLMs) have excelled in processing multiple languages through pre- and post-tr…
用问答任务探测大模型隐藏的认知盲区，揭秘AI到底“知道什么却不说”
arXiv:2605.31561v1 Announce Type: new Abstract: Test-time reasoning has become a significant field of study since the introduction of chain-of-thought…
论文揭示了LLM在回答时隐藏的“盲点偏见”，帮你发现模型没说什么，提升AI对齐与安全认知。
arXiv:2602.10117v5 Announce Type: replace-cross Abstract: Large Language Models (LLMs) often provide chain-of-thought (CoT) reasoning traces that appe…
揭秘思维链内部机制，从解码、投影、激活三角度追踪信息流，ACL 2026收录。
arXiv:2507.20758v2 Announce Type: replace Abstract: Chain-of-Thought (CoT) prompting significantly enhances model reasoning, yet its internal mechanis…
反事实知识训练虽能帮助LLM遗忘特定信息，但论文揭示其背后的隐藏代价，值得关注。
arXiv:2605.27083v1 Announce Type: new Abstract: Counterfactual tuning (CFT) has emerged as a promising paradigm for Large Language Model (LLM) unlearn…
知识图谱作为关键数据层，补齐LLM在工业资产管理中的结构化短板，实现更可靠的操作决策。
arXiv:2605.26874v1 Announce Type: cross Abstract: LLM-based agents for industrial asset operations show limited accuracy when reasoning over flat docu…
用少量提示即可精准定位大模型中的关键神经元，揭示AI“大脑”的运作秘密。
arXiv:2605.24846v1 Announce Type: cross Abstract: Large language models (LLMs) display strong comprehensive abilities, yet the internal mechanisms tha…
通过激活修补技术探究LLM遗忘的深浅程度，为模型安全对齐提供新测量维度
arXiv:2605.24614v1 Announce Type: cross Abstract: Large language model (LLM) unlearning has emerged as a crucial post-hoc mechanism for privacy protec…