Regularized Best-of-N Sampling to Mitigate Reward Hacking for Language Model Alignment

Abstract

Best-of-N (BoN) sampling with a reward model has been shown to be an effective strategy for aligning Large Language Models (LLMs) to human preferences at the time of decoding. BoN sampling is susceptible to a problem known as \textit{reward hacking}. Because the reward model is an imperfect proxy for the true objective, over-optimizing its value can compromise its performance on the true objective. A common solution to prevent reward hacking in preference learning techniques is to optimize a reward using proximity regularization (e.g., KL regularization), which ensures that the language model remains close to the reference model. In this research, we propose Regularized Best-of-N (RBoN), a variant of BoN that aims to mitigate reward hacking by incorporating a proximity term in response selection, similar to preference learning techniques. We evaluate RBoN on the AlpacaFarm and Anthropic's hh-rlhf datasets and show that it outperforms BoN. As an application of RBoN, we use RBoN to generate a pairwise preference learning dataset. Experimental results show that a DPO model trained on a dataset generated with RBoN outperforms a DPO model generated with vanilla BoN.