# Policy Gradient Primal-Dual Method for Safe Reinforcement Learning from Human Feedback

arXiv: https://arxiv.org/abs/2604.19024 · HTML: https://mlsanity.com/paper/2604.19024/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 26 of its 38 references that are in the map.
- First posted on arXiv 21 Apr 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Reward modeling and RLHF](https://mlsanity.com/alignment/reward-modeling-and-rlhf/)

## Builds on (most cited first, 26 of 26)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [Constrained Policy Optimization](https://mlsanity.com/paper/1705.10528/) (2017)
- [Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2307.15217/) (2023)
- [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://mlsanity.com/paper/2209.07858/) (2022)
- [Safe RLHF: Safe Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.12773/) (2023)
- [Reward Constrained Policy Optimization](https://mlsanity.com/paper/1805.11074/) (2018)
- [Risk-Constrained Reinforcement Learning with Percentile Risk Criteria](https://mlsanity.com/paper/1512.01629/) (2015)
- [Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision](https://mlsanity.com/paper/2305.03047/) (2023)
- [A Survey of Inverse Reinforcement Learning: Challenges, Methods and Progress](https://mlsanity.com/paper/1806.06877/) (2018)
- [Principled Reinforcement Learning with Human Feedback from Pairwise or K-wise Comparisons](https://mlsanity.com/paper/2301.11270/) (2023)
- [Constrained Reinforcement Learning Has Zero Duality Gap](https://mlsanity.com/paper/1910.13393/) (2019)
- [Zeroth-Order Stochastic Variance Reduction for Nonconvex Optimization](https://mlsanity.com/paper/1805.10367/) (2018)
- [Human-in-the-loop: Provably Efficient Preference-based Reinforcement Learning with General Function Approximation](https://mlsanity.com/paper/2205.11140/) (2022)
- [Provable Offline Preference-Based Reinforcement Learning](https://mlsanity.com/paper/2305.14816/) (2023)
- [Making RL with Preference-based Feedback Efficient via Randomization](https://mlsanity.com/paper/2310.14554/) (2023)
- [A Sample-Efficient Algorithm for Episodic Finite-Horizon MDP with Constraints](https://mlsanity.com/paper/2009.11348/) (2020)
- [On the Theory of Reinforcement Learning with Once-per-Episode Feedback](https://mlsanity.com/paper/2105.14363/) (2021)
- [Convergence and sample complexity of natural policy gradient primal-dual methods for constrained MDPs](https://mlsanity.com/paper/2206.02346/) (2022)
- [Provable Reward-Agnostic Preference-Based Reinforcement Learning](https://mlsanity.com/paper/2305.18505/) (2023)
- [Exploration-Driven Policy Optimization in RLHF: Theoretical Insights on Efficient Data Utilization](https://mlsanity.com/paper/2402.10342/) (2024)
- [Zeroth-Order Policy Gradient for Reinforcement Learning from Human Feedback without Reward Inference](https://mlsanity.com/paper/2409.17401/) (2024)
- [Reinforcement Learning from Human Feedback without Reward Inference: Model-Free Algorithm and Instance-Dependent Analysis](https://mlsanity.com/paper/2406.07455/) (2024)

## Built on by (most cited first, 2 of 2)

- [Beyond Uniform Forgetting: A Study of Sequential Direct Preference Optimization Across Preference Settings](https://mlsanity.com/paper/2606.19744/) (2026)
- [Understanding the Impact of AI Code Assistants on Security API Usage: An Empirical Study](https://mlsanity.com/paper/2607.11348/) (2026)

## Abstract (arXiv)

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.