2025
ACL
ACL 2025
Atyaephyra at SemEval-2025 Task 4: Low-Rank Negative Preference Optimization
Abstract
AbstractWe present a submission to the SemEval 2025 shared task on unlearning sensitive content from LLMs. Our approach employs negative preference optimization using low-rank adaptation. We show that we can utilize this combination to cheaply compute additional regularization terms, which help with unlearning stabilization. The results of our approach significantly exceed the shared task baselines.
🌉
Interdisciplinary Bridge
— Artificial Intelligence and Deep Learning and Machine Learning and Natural Language Processing
🧭
Keyword Pioneer
— negative preference optimization
🐝
Cross-Pollinator
— Artificial Intelligence, Computer Science, Computer Vision, Data Science & Analytics, Deep Learning, Healthcare & Medicine, Interdisciplinary, Knowledge & Reasoning, Machine Learning, Mathematics & Optimization, Natural Language Processing, Reinforcement Learning, Robotics, Security & Privacy, Speech & Audio
Authors
Topics
Artificial Intelligence > Core AI > AI Safety
Artificial Intelligence > Core AI > Model Compression
Artificial Intelligence > Core AI > Responsible AI
Machine Learning > Optimization & Theory > Optimization
Machine Learning > Application Areas > Privacy
Deep Learning > Models > Generative Models
Deep Learning > Techniques > Model Architecture
Natural Language Processing > Resources & Methods > Large Language Models