Detecting Conceptual Abstraction in LLMs

arXiv:2404.15848v1 Announce Type: new
Abstract: We present a novel approach to detecting noun abstraction within a large language model (LLM). Starting from a psychologically motivated set of noun pairs in taxonomic relationships, we instantiate surface patterns indicating hypernymy and analyze the attention matrices produced by BERT. We compare the results to two sets of counterfactuals and show that we can detect hypernymy in the abstraction mechanism, which cannot solely be related to the distributional similarity of noun pairs. Our findings …

abstract abstraction analyze arxiv attention bert cs.cl cs.lg language language model large language large language model llm llms novel patterns relationships results set surface type