2024/06/20 by Mandar Sharma, Sharma, Mandar, Nikhil Muralidhar +6
Computer Science · #Artificial Intelligence (cs.AI) #Artificial intelligence #Computation and Language (cs.CL) #Computer science #FOS: Computer and information sciences #Machine Learning (cs.LG) #Natural Language Processing Techniques #Natural language processing #Regularization (linguistics) #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2406.14005
published in arXiv (Cornell University) (Cornell University)
openalex publication_date 2024/06/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The pretraining-fine-tuning paradigm has been the de facto strategy for transfer learning in modern language modeling. With the understanding that task adaptation in LMs is often a function of parameters shared across tasks, we argue that a more surgical approach to regularization needs to exist for smoother transfer learning. Towards this end, we investigate how the pretraining loss landscape is affected by these task-sensitive parameters through an information-theoretic lens. We then leverage the findings from our investigations to devise a novel approach to dropout for improved model regularization and better downstream generalization. This approach, named guided dropout, is both task & architecture agnostic and adds no computational overhead to the fine-tuning process. Through empirical evaluations, we showcase that our approach to regularization yields consistently better performance, even in scenarios of data paucity, compared to standardized baselines.