What GPT Knows About Who is Who
This addresses the challenge of expensive supervised labels in coreference resolution for NLP researchers, though it is incremental as it primarily evaluates existing models.
The paper tackled the problem of coreference resolution using large language models (LLMs) through a QA-based prompt-engineering method, finding that GPT-2 and GPT-Neo can return valid answers but have limited and inconsistent capabilities for identifying coreferent mentions.
Coreference resolution -- which is a crucial task for understanding discourse and language at large -- has yet to witness widespread benefits from large language models (LLMs). Moreover, coreference resolution systems largely rely on supervised labels, which are highly expensive and difficult to annotate, thus making it ripe for prompt engineering. In this paper, we introduce a QA-based prompt-engineering method and discern \textit{generative}, pre-trained LLMs' abilities and limitations toward the task of coreference resolution. Our experiments show that GPT-2 and GPT-Neo can return valid answers, but that their capabilities to identify coreferent mentions are limited and prompt-sensitive, leading to inconsistent results.