Key points are not available for this paper at this time.
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.
Building similarity graph...
Analyzing shared references across papers
Loading...
Sun et al. (Fri,) studied this question.
www.synapsesocial.com/papers/68e68995b6db6435876121c5 — DOI: https://doi.org/10.48550/arxiv.2405.15624
Hao Sun
Mihaela van der Schaar
Building similarity graph...
Analyzing shared references across papers
Loading...