1
PRISM Edit: One Vector for All Temporal Answers
单一向量让大模型同时记住新旧答案，攻克时间事实编辑难题
arXiv:2607.11327v1 Announce Type: cross Abstract: Model editing keeps large language models (LLMs) up to date without retraining, but temporal facts e…
2
CBD: API-Only LLM Black-Box Unlearning through Controlled Behavioral Divergence
仅通过API接口实现LLM黑盒遗忘，用行为分歧控制精准擦除特定知识，模型编辑与隐私保护新方向。
arXiv:2606.27683v1 Announce Type: cross Abstract: Edge devices increasingly invoke large language models (LLMs) through API services for context aware…
3
Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLMs
多模态大语言模型在配对时表现准确、拆分时出错，研究通过解耦编辑特定模态神经元来修复这一现象。
arXiv:2606.17057v1 Announce Type: cross Abstract: Although Knowledge Editing provides an efficient mechanism for updating the knowledge of Multimodal …
4
BrainSurgery: Reproducible and Reliable Declarative Weight Manipulations for Model Editing and Upcycling
可复现且可靠的声明式模型权重操作方法，为AI模型编辑与升级提供新范式
arXiv:2606.09707v1 Announce Type: new Abstract: As deep learning models scale, managing, inspecting, and modifying large checkpoints has become increa…
5
Evaluating the Reversal Curse in Model Editing
一篇评估大模型编辑中「反转诅咒」现象的论文，揭示知识修改后的对称性失效问题。
arXiv:2310.10322v3 Announce Type: replace Abstract: Large language models (LLMs) are prone to hallucinate unintended text due to false or outdated kno…
6
Reverse-Engineering Model Editing on Language Models
ICML 2026论文揭秘：如何逆向工程语言模型的编辑操作，深入理解模型内部机制。
arXiv:2602.10134v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are pretrained on corpora containing trillions of tokens and, t…