Can RLHF be More Efficient with Imperfect Reward Models? A Policy Coverage Perspective.

Jiawei Huang, Bingcong Li, Christoph Dann, Niao He

Browse the full ICML paper archive.