Unlearning Traces the Influential Training Data of Language Models

arXiv:2401.15241v2 Announce Type: replace
Abstract: Identifying the training datasets that influence a language model's outputs is essential for minimizing the generation of harmful content and enhancing its performance. Ideally, we can measure the influence of each dataset by removing it from training; however, it is prohibitively expensive to retrain a model multiple times. This paper presents UnTrac: unlearning traces the influence of a training dataset on the model's performance. UnTrac is extremely simple; each training dataset is unlearned by gradient …

abstract arxiv cs.ai cs.cl data dataset datasets however influence language language model language models performance replace retrain traces training training data training datasets type unlearning