#Anchor-Grounded Risk

1 stories · 0 in 7 days · 1 in 30 days

entity-evidence binding 1 Fixed-Anchor Correction 1 LLM agents 1 Phantom Merge 1

Stories, products and papers tagged "Anchor-Grounded Risk", newest first.

10月15日

Papers Builders00:37

Chinese contentPhantom Merge：大语言模型实体绑定错误研究

研究人员发现LLM代理常将错误主张归因于错误实体，他们开发的检测方法准确率达96.2%，修正方法将错误率降低88%。

#Phantom Merge #LLM agents #entity-evidence binding #Anchor-Grounded Risk

arXiv cs.AI@Wangshu Zhu et al. (3)Source

Anchor-Grounded Risk · AI news | AITOP