Recite, Reconstruct, Recollect: Memorization in LMs as a Multifaceted Phenomenon
Microsoft, Prev EleutherAI · DatologyAI EleutherAI · Microsoft · New York University · MPI-SWS · Northeastern University · Google DeepMind · University of Illinois at Urbana-Champaign · EleutherAI · Harvard University · Cornell University · Kempner Institute, Harvard University
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
Memorization in language models is typically treated as a homogenous phenomenon, neglecting the specifics of the memorized data. We instead model memorization as the effect of a set of complex factors that describe each sample and relate it to the model and corpus. To build intuition around these factors, we break memorization down into a taxonomy: recitation of highly duplicated sequences, reconstruction of inherently predictable sequences, and recollection of sequences that are neither. We demonstrate the usefulness of our taxonomy by using it to construct a predictive model for memorization. By analyzing dependencies and inspecting the weights of the predictive model, we find that different factors have different influences on the likelihood of memorization depending on the taxonomic category.