Inverse-RLignment: Inverse Reinforcement Learning from Demonstrations for LLM Alignment
Sun, Hao, van der Schaar, Mihaela
–arXiv.org Artificial Intelligence
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.
arXiv.org Artificial Intelligence
May-24-2024
- Country:
- North America > United States
- Illinois > Cook County > Chicago (0.04)
- Europe > United Kingdom
- England > Cambridgeshire > Cambridge (0.04)
- North America > United States
- Genre:
- Research Report > New Finding (1.00)
- Industry:
- Leisure & Entertainment > Games (1.00)
- Information Technology > Security & Privacy (0.66)
- Technology: