大规模模型编辑导致渐进性与灾难性遗忘
Model Editing at Scale leads to Gradual and Catastrophic Forgetting
AI总结：
本文大规模评估了ROME和MEMIT两种模型编辑方法，发现多次顺序编辑会导致渐进性和灾难性遗忘，从而限制模型编辑的实用性和可扩展性，并呼吁在评估中重视可扩展性。
摘要：
在大语言模型中编辑知识是一种极具吸引力的能力，它使我们能够纠正预训练期间学习到的错误事实，并用不断增长的新事实列表更新模型。虽然现有的模型编辑技术已展现出潜力，但它们的评估通常使用可靠性、特异性和泛化性指标，且仅针对一次或少数几次编辑。我们认为，要使模型编辑具有实际应用价值，我们必须能够对同一模型进行多次编辑。基于这一考虑，我们对当前的模型编辑方法进行了大规模评估，重点关注两种最先进的方法：ROME和MEMIT。我们发现，当模型被顺序编辑多个事实时，它会持续遗忘先前编辑过的事实以及执行下游任务的能力。这种遗忘分为两个阶段——首先是渐进的但持续加剧的遗忘阶段，随后是突然的或灾难性的遗忘阶段。渐进性遗忘和灾难性遗忘都限制了模型编辑方法在大规模场景下的实用性——前者使得对模型进行多次编辑时模型编辑的效果逐渐降低，而后者则限制了此类模型编辑方法的可扩展性上限。我们的分析还揭示了ROME和MEMIT在大规模场景下的其他关键局限性。通过我们的工作，我们推动在考虑可扩展性的前提下开发和评估模型编辑方法。
英文摘要：
Editing knowledge in large language models is an attractive capability to have which allows us to correct incorrectly learnt facts during pre-training, as well as update the model with an ever-growing list of new facts. While existing model editing techniques have shown promise, they are usually evaluated using metrics for reliability, specificity and generalization over one or few edits. We argue that for model editing to have practical utility, we must be able to make multiple edits to the same model. With this in mind, we evaluate the current model editing methods at scale, focusing on two state of the art methods: ROME and MEMIT. We find that as the model is edited sequentially with multiple facts, it continually forgets previously edited facts and the ability to perform downstream tasks. This forgetting happens in two phases -- an initial gradual but progressive forgetting phase followed by abrupt or catastrophic forgetting phase. Both gradual and catastrophic forgetting limit the usefulness of model editing methods at scale -- the former making model editing less effective as multiple edits are made to the model while the latter caps the scalability of such model editing methods. Our analysis also highlights other key limitations of ROME and MEMIT at scale. With our work, we push for the development and evaluation of model editing methods keeping scalability in mind.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
