conftrace_

online rlhf

2 papers

Explore in graph

Co-occurring keywords · papers mentioning both

human preference sampling (1) iterative rlhf (1) oracle reward (1) pilaf (1) win-rate-based selection (1) suboptimality (1) policy coverage (1) kl regularization (1) transfer learning (1) reward modeling (1)

Papers

Can RLHF be More Efficient with Imperfect Reward Models? A Policy Coverage Perspective ICML 2025 PILAF: Optimal Human Preference Sampling for Reward Modeling ICML 2025