Can I read Ferrando 等 - 2025 - Do i Know This Entity Knowledge Awareness and Hallucinations in Language Models on EtoBox?

Ferrando 等 - 2025 - Do i Know This Entity Knowledge Awareness and Hallucinations in Language Models by 486763647 is a document available to read on EtoBox.

What is Ferrando 等 - 2025 - Do i Know This Entity Knowledge Awareness and Hallucinations in Language Models about?

This paper investigates the mechanisms behind hallucinations in large language models (LLMs), focusing on entity recognition and the concept of knowledge awareness. Using sparse autoencoders, the authors identify directions in the representation space that indicate whether a model can recall facts about entities, which can influence its tendency to hallucinate or refuse to answer questions. The findings suggest that these entity recognition mechanisms are causally relevant and can be manipulated to affect t

Author

486763647

Language

EN