English
Related papers

Related papers: Pluralistic Off-policy Evaluation and Alignment

200 papers

The problem of Offline Policy Evaluation (OPE) in Reinforcement Learning (RL) is a critical step towards applying RL in real-life applications. Existing work on OPE mostly focus on evaluating a fixed target policy $\pi$, which does not…

Machine Learning · Computer Science 2020-12-02 Ming Yin , Yu Bai , Yu-Xiang Wang

Learning a reward function from human preferences is challenging as it typically requires having a high-fidelity simulator or using expensive and potentially unsafe actual physical rollouts in the environment. However, in many tasks the…

Machine Learning · Computer Science 2023-01-05 Daniel Shin , Anca D. Dragan , Daniel S. Brown

Current alignment pipelines presume a single, universal notion of desirable behavior. However, human preferences often diverge across users, contexts, and cultures. As a result, disagreement collapses into the majority signal and minority…

Machine Learning · Computer Science 2025-06-10 Daniel Halpern , Evi Micha , Ariel D. Procaccia , Itai Shapira

Preference learning is a key technology for aligning language models with human values. Reinforcement Learning from Human Feedback (RLHF) is a model-based algorithm to optimize preference learning, which first fits a reward model for…

Machine Learning · Computer Science 2024-03-26 Zaifan Jiang , Xing Huang , Chao Wei

We offer an experimental benchmark and empirical study for off-policy policy evaluation (OPE) in reinforcement learning, which is a key problem in many safety critical applications. Given the increasing interest in deploying learning-based…

Machine Learning · Computer Science 2021-11-30 Cameron Voloshin , Hoang M. Le , Nan Jiang , Yisong Yue

We study minimax methods for off-policy evaluation (OPE) using value functions and marginalized importance weights. Despite that they hold promises of overcoming the exponential variance in traditional importance sampling, several key…

Machine Learning · Computer Science 2020-11-06 Nan Jiang , Jiawei Huang

Aligning with human preference datasets has been critical to the success of large language models (LLMs). Reinforcement learning from human feedback (RLHF) employs a costly reward model to provide feedback for on-policy sampling responses.…

Machine Learning · Computer Science 2024-05-24 Yuanzhao Zhai , Zhuo Zhang , Kele Xu , Hanyang Peng , Yue Yu , Dawei Feng , Cheng Yang , Bo Ding , Huaimin Wang

We consider evaluating and training a new policy for the evaluation data by using the historical data obtained from a different policy. The goal of off-policy evaluation (OPE) is to estimate the expected reward of a new policy over the…

Machine Learning · Statistics 2020-10-19 Masahiro Kato , Masatoshi Uehara , Shota Yasui

Off-policy evaluation (OPE) is a method for estimating the return of a target policy using some pre-collected observational data generated by a potentially different behavior policy. In some cases, there may be unmeasured variables that can…

Machine Learning · Statistics 2023-02-03 Yang Xu , Jin Zhu , Chengchun Shi , Shikai Luo , Rui Song

In this paper, we investigate preference-based reinforcement learning (PbRL), which enables reinforcement learning (RL) agents to learn from human feedback. This is particularly valuable when defining a fine-grain reward function is not…

Machine Learning · Computer Science 2025-11-11 Guojian Wang , Jianxiang Liu , Xinyuan Li , Faguo Wu , Xiao Zhang , Tianyuan Chen , Xuyang Chen

While astonishingly capable, large Language Models (LLM) can sometimes produce outputs that deviate from human expectations. Such deviations necessitate an alignment phase to prevent disseminating untruthful, toxic, or biased information.…

Artificial Intelligence · Computer Science 2024-10-30 Long Tan Le , Han Shu , Tung-Anh Nguyen , Choong Seon Hong , Nguyen H. Tran

Bootstrapping large language models (LLMs) through preference-based policy optimization offers a promising direction for aligning model behavior with human preferences without relying on extensive manual annotations. In this work, we…

Artificial Intelligence · Computer Science 2025-12-25 Chen Jia

Supervised fine-tuning (SFT) has emerged as a crucial method for aligning large language models (LLMs) with human-annotated demonstrations. However, SFT, being an off-policy approach similar to behavior cloning, often struggles with…

Computation and Language · Computer Science 2025-10-27 Qingru Zhang , Liang Qiu , Ilgee Hong , Zhenghao Xu , Tianyi Liu , Shiyang Li , Rongzhi Zhang , Zheng Li , Lihong Li , Bing Yin , Chao Zhang , Jianshu Chen , Haoming Jiang , Tuo Zhao

Reinforcement learning (RL) is increasingly used to align large language models (LLMs). Off-policy methods offer greater implementation simplicity and data efficiency than on-policy techniques, but often result in suboptimal performance. In…

Machine Learning · Computer Science 2025-12-01 Charles Arnal , Gaëtan Narozniak , Vivien Cabannes , Yunhao Tang , Julia Kempe , Remi Munos

In this paper, we investigate the problem of offline Preference-based Reinforcement Learning (PbRL) with human feedback where feedback is available in the form of preference between trajectory pairs rather than explicit rewards. Our…

Machine Learning · Computer Science 2023-10-03 Wenhao Zhan , Masatoshi Uehara , Nathan Kallus , Jason D. Lee , Wen Sun

Motivated by the many real-world applications of reinforcement learning (RL) that require safe-policy iterations, we consider the problem of off-policy evaluation (OPE) -- the problem of evaluating a new policy using the historical data…

Machine Learning · Computer Science 2020-04-02 Tengyang Xie , Yifei Ma , Yu-Xiang Wang

Alignment of large language models (LLMs) typically involves training a reward model on preference data, followed by policy optimization with respect to the reward model. However, optimizing policies with respect to a single reward model…

Machine Learning · Computer Science 2025-07-23 Debangshu Banerjee , Kintan Saha , Aditya Gopalan

Reinforcement Learning from Human Feedback (RLHF) allows us to train models, such as language models (LMs), to follow complex human preferences. In RLHF for LMs, we first train an LM using supervised fine-tuning, sample pairs of responses,…

Machine Learning · Computer Science 2025-07-22 Johannes Ackermann , Takashi Ishida , Masashi Sugiyama

Holdout validation and hyperparameter tuning from data is a long-standing problem in offline reinforcement learning (RL). A standard framework is to use off-policy evaluation (OPE) methods to evaluate and select the policies, but OPE either…

Machine Learning · Computer Science 2025-10-27 Pai Liu , Lingfeng Zhao , Shivangi Agarwal , Jinghan Liu , Audrey Huang , Philip Amortila , Nan Jiang

Reinforcement learning (RL) approaches for Large Language Models (LLMs) frequently use on-policy algorithms, such as PPO or GRPO. However, policy lag from distributed training architectures and differences between the training and inference…

Machine Learning · Computer Science 2026-03-03 Daniel Ritter , Owen Oertell , Bradley Guo , Jonathan Chang , Kianté Brantley , Wen Sun