Search papers, labs, and topics across Lattice.
This paper formulates Safe RLHF as an infinite horizon discounted Constrained Markov Decision Process (CMDP) to better reflect real-world human-AI interaction. They introduce two primal-dual policy gradient algorithms that directly optimize for helpfulness and harmlessness without requiring reward model fitting. The algorithms achieve global convergence guarantees with polynomial rates, marking the first such result for infinite horizon discounted CMDPs under human feedback.
Forget reward model fitting: these primal-dual policy gradient methods offer provably safe and convergent RLHF in infinite horizon settings.
Safe Reinforcement Learning from Human Feedback (Safe RLHF) has recently achieved empirical success in developing helpful and harmless large language models by decoupling human preferences regarding helpfulness and harmlessness. Existing approaches typically rely on fitting fixed horizon reward models from human feedback and have only been validated empirically. In this paper, we formulate safe RLHF as an infinite horizon discounted Con- strained Markov Decision Process (CMDP), since humans may interact with the model over a continuing sequence of interactions rather than within a single finite episode. We propose two Safe RLHF algorithms that do not require reward model fitting and, in contrast to prior work assuming fixed-length trajectories, support flexible trajectory lengths for training. Both algo- rithms are based on the primal-dual method and achieve global convergence guarantees with polynomial rates in terms of policy gradient iterations, trajectory sample lengths, and human preference queries. To the best of our knowledge, this is the first work to study infinite horizon discounted CMDP under human feedback and establish global, non-asymptotic convergence.
