DocMEdit: Towards Document-Level Model Editing
This work addresses the problem of making model editing more practical for real-world document-level applications, though it is incremental as it extends existing tasks to a new scope.
The authors tackled the limitation of existing model editing datasets, which focus on short outputs, by proposing document-level model editing to address real-world tasks, and introduced a new dataset showing that current methods struggle with this challenge.
Model editing aims to correct errors and outdated knowledge in the Large language models (LLMs) with minimal cost. Prior research has proposed a variety of datasets to assess the effectiveness of these model editing methods. However, most existing datasets only require models to output short phrases or sentences, overlooks the widespread existence of document-level tasks in the real world, raising doubts about their practical usability. Aimed at addressing this limitation and promoting the application of model editing in real-world scenarios, we propose the task of document-level model editing. To tackle such challenges and enhance model capabilities in practical settings, we introduce \benchmarkname, a dataset focused on document-level model editing, characterized by document-level inputs and outputs, extrapolative, and multiple facts within a single edit. We propose a series of evaluation metrics and experiments. The results show that the difficulties in document-level model editing pose challenges for existing model editing methods.