A Controlled Reevaluation of Coreference Resolution Models

arXiv:2404.00727v1 Announce Type: new
Abstract: All state-of-the-art coreference resolution (CR) models involve finetuning a pretrained language model. Whether the superior performance of one CR model over another is due to the choice of language model or other factors, such as the task-specific architecture, is difficult or impossible to determine due to lack of a standardized experimental setup. To resolve this ambiguity, we systematically evaluate five CR models and control for certain design decisions including the pretrained language model used by …

abstract architecture art arxiv cs.cl finetuning language language model performance pretrained language model resolution state type