2026
Human-AI Curation Synergy: Scaling Preference Data Curation via Human-Guided AI Feedback
ICLR 2026poster
Despite the critical role of reward models (RMs) in reinforcement learning from human feedback (RLHF), current state-of-the-art open RMs perform poorly on most existing evaluation benchmarks, failing to capture the spectrum of nuanced and sophisticated human preferences. Even approaches incorporatin…