Large Language Model Alignment via Inverse Reinforcement Learning from Demonstrations
Keywords: Large Language Model Alignment, Alignment from Demonstration
Abstract: Aligning Large Language Models (LLMs) is crucial for enhancing their safety and utility. However, existing methods, primarily based on preference datasets, face challenges such as noisy labels, high annotation costs, and privacy concerns.
In this work, we introduce **_Alignment from Demonstrations_** (AfD), a novel approach leveraging high-quality demonstration data to overcome these challenges. We formalize AfD within a sequential decision-making framework, highlighting its unique challenge of missing reward signals. Drawing insights from forward and inverse reinforcement learning, we introduce divergence minimization objectives for AfD.
Analytically, we elucidate the mass-covering and mode-seeking behaviors of various approaches, explaining when and why certain methods are superior.
Practically, we propose a computationally efficient algorithm that extrapolates over a tailored reward model for AfD. We validate our key insights through experiments on the Harmless and Helpful tasks, demonstrating their strong empirical performance while maintaining simplicity.
Primary Area: alignment, fairness, safety, privacy, and societal considerations
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 7805
Loading