共计 246 篇文章
2026
从记忆视角，你也能发明 Transformer
2026年多模态大模型幻觉（Multimodal LLM Hallucination）论文精读
ThinkTrap： Denial-of-Service Attacks against Black-box LLM Services via Infinite Thinking
记录：第一届中国网络空间安全大会
Purifying Generative LLMs from Backdoors without Prior Knowledge or Clean Reference
Harnessing the Universal Geometry of Embeddings
Constructing Unlearnable Data with Solely Linear Classifiers
Understanding and Mitigating Hallucinations in Multimodal Chain-of-Thought Models
Small LLMs Are Weak Tool Learners： A Multi-LLM Agen
KVSmooth： Mitigating Hallucination in Multi-modal Large Language Models through Key-Value Smoothing