共计 116 篇文章

---

2025

09-28

通过外推一个巨大且假设的语言模型的概率来解释和改进对比解码

09-27

五篇幻觉相关论文速览

09-26

Machine Unlearning

09-19

MokA：Multimodal Low-Rank Adaptation for MLLMs

09-17

Mitigating Hallucinations in Large Vision-Language Models by Self-Injecting Hallucinations

09-15

Nullu：通过 HalluSpace 投影减轻大型视觉-语言模型中的对象幻觉

09-14

Vision Transformers Don't Need Trained Registers

09-13

LLM中MOE的安全行为

09-10

DoLa：通过对比层解码提高大型语言模型的事实性

09-08

PerturboLLaVA：通过扰动视觉训练减少多模态幻觉