Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking

arXiv:2402.14811v1 Announce Type: cross
Abstract: Fine-tuning on generalized tasks such as instruction following, code generation, and mathematics has been shown to enhance language models' performance on a range of tasks. Nevertheless, explanations of how such fine-tuning influences the internal computations in these models remain elusive. We study how fine-tuning affects the internal mechanisms implemented in language models. As a case study, we explore the property of entity tracking, a crucial facet of language comprehension, where models fine-tuned on mathematics have …

abstract arxiv case case study code code generation cs.cl cs.lg fine-tuning generalized language language models mathematics performance study tasks tracking type