Reverse-Kullback-Leibler (KL) regularization has emerged to be a predominant technique to enhance policy optimization in reinforcement learning (RL) and reinforcement learning from human feedback (RLHF), which forces the learned policy to stay close to a reference policy. While the effectiveness of KL-regularization has been empirically demonstrated in various practical scenarios, current theoretical analyses of KL-regularized RLHF still yield the same O(1/ϵ2) sample complexity as ones without KL-regularization. To understand the fundamental distinction between objectives with KL-regularization and ones without KLregularization, we are the first to theoretically demonstrate the power of KLregularization by providing a sharp analysis for KL-regularized contextual bandits and RLHF, revealing an O(1/ϵ) sample complexity when ϵ is sufficiently small. We also prove matching lower bounds for both settings. More specifically, we study how the coverage of the reference policy affects the sample complexity of KL-regularized online contextual bandits and RLHF. We show that with sufficient coverage from the reference policy, a simple two-stage mixed sampling algorithm can achieve an O(1/ϵ) sample complexity with only an additive dependence on the coverage coefficient, thus proving the benefits of online data even without explicit exploration. Our results provide a comprehensive understanding of the roles of KL-regularization and data coverage in online decision making, shedding light on the design of more efficient algorithms.

Human feedback often arrives as preferences rather than calibrated numeric rewards, motivating reinforcement learning from preferential feedback, also referred to as reinforcement learning from human feedback (RLHF). We present a rigorous theoretical study of preference-only learning in episodic kernel MDPs. In each episode, the learner deploys two policies from a common start state and receives a single binary label indicating which trajectory is preferred, modeled by a Bradley--Terry--Luce link on the difference of cumulative (unobserved) rewards. Under kernel-based assumptions on the reward and transition functions (one of the most general models amenable to theoretical analysis) we develop preference-based value estimation and confidence sets tailored to end-of-episode comparisons. We prove high-probability regret bounds that scale sublinearly in the number of episodes, implying that the value of the learned policy converges to that of the optimal policy.

Preference feedback, in the form of pairwise comparisons rather than scalar scores, has seen increasing use in applications such as human-, laboratory-, and expert-in-the-loop design, as well as scientific discovery. We propose a Thompson Sampling (TS) approach to Bayesian optimization with preferential feedback that models comparisons using a monotone link on latent utility differences and leverages the dueling kernel induced by a base kernel. We provide a finite-time analysis showing that the performance of the proposed method matches that of standard TS for conventional Bayesian optimization with scalar feedback. The analysis exploits the anchor invariance of TS for challenger selection and introduces a double-TS pairing variant. We also demonstrate the performance of the method on both synthetic and real-world examples.

Learning from preference feedback has emerged as an essential step for improving the generation quality and performance of modern language models (LMs). Despite its widespread use, the way preference-based learning is applied varies wildly, with differing data, learning algorithms, and evaluations used, making disentangling the impact of each aspect difficult. In this work, we identify four core aspects of preference-based learning: preference data, learning algorithm, reward model, and policy training prompts, systematically investigate the impact of these components on downstream model performance, and suggest a recipe for strong learning for preference feedback. Our findings indicate that all aspects are important for performance, with better preference data leading to the largest improvements, followed by the choice of learning algorithm, the use of improved reward models, and finally the use of additional unlabeled prompts for policy training. Notably, PPO outperforms DPO by up to 2.5% in math and 1.2% in general domains. High-quality preference data leads to improvements of up to 8% in instruction following and truthfulness. Despite significant gains of up to 5% in mathematical evaluation when scaling up reward models, we surprisingly observe marginal improvements in other categories.

Neural Information Processing Systems http://nips.cc/

Neural Information Processing Systems http://nips.cc/