English
Related papers

Related papers: Detecting and Suppressing Reward Hacking with Grad…

200 papers

While Supervised Fine-Tuning (SFT) and Rejection Sampling Fine-Tuning (RFT) are standard for LLM alignment, they either rely on costly expert data or discard valuable negative samples, leading to data inefficiency. To address this, we…

Machine Learning · Computer Science 2026-04-24 Zehua Liu , Shuqi Liu , Tao Zhong , Mingxuan Yuan

Recent RL methods have substantially improved the reasoning abilities of LLMs. Existing reward designs mainly follow two paradigms: (1) Reinforcement learning with verifiable rewards (RLVR) derives outcome signals from executable checks or…

Computation and Language · Computer Science 2026-05-25 Sirui Chen , Lei Xu , Yuying Zhao , Yutian Chen , Yu Wang , Beier Zhu , Hanwang Zhang , Shengjie Zhao , Chaochao Lu

Reinforcement learning from human feedback (RLHF) has become a powerful post-training paradigm for aligning large language models with human preferences. A core challenge in RLHF is constructing accurate reward signals, where the…

Machine Learning · Computer Science 2025-05-23 Ilgee Hong , Changlong Yu , Liang Qiu , Weixiang Yan , Zhenghao Xu , Haoming Jiang , Qingru Zhang , Qin Lu , Xin Liu , Chao Zhang , Tuo Zhao

We explore how iterative revising a chain of thoughts with the help of information retrieval significantly improves large language models' reasoning and generation ability in long-horizon generation tasks, while hugely mitigating…

Computation and Language · Computer Science 2024-03-11 Zihao Wang , Anji Liu , Haowei Lin , Jiaqi Li , Xiaojian Ma , Yitao Liang

Reward function design and exploration time are arguably the biggest obstacles to the deployment of reinforcement learning (RL) agents in the real world. In many real-world tasks, designing a reward function takes considerable hand…

Computer Vision and Pattern Recognition · Computer Science 2017-06-14 Pierre Sermanet , Kelvin Xu , Sergey Levine

Step-by-step verifiers -- also known as process reward models (PRMs) -- are a key ingredient for test-time scaling. PRMs require step-level supervision, making them expensive to train. This work aims to build data-efficient PRMs as…

Machine Learning · Computer Science 2025-12-09 Muhammad Khalifa , Rishabh Agarwal , Lajanugen Logeswaran , Jaekyeom Kim , Hao Peng , Moontae Lee , Honglak Lee , Lu Wang

Large reasoning models (LRMs) have recently shown promise in solving complex math problems when optimized with Reinforcement Learning (RL). But conventional approaches rely on outcome-only rewards that provide sparse feedback, resulting in…

Machine Learning · Computer Science 2025-08-01 Tao He , Rongchuan Mu , Lizi Liao , Yixin Cao , Ming Liu , Bing Qin

Recent advances in multimodal Reward Models (RMs) have shown significant promise in delivering reward signals to align vision models with human preferences. However, current RMs are generally restricted to providing direct responses or…

Computer Vision and Pattern Recognition · Computer Science 2025-10-30 Yibin Wang , Zhimin Li , Yuhang Zang , Chunyu Wang , Qinglin Lu , Cheng Jin , Jiaqi Wang

While Reinforcement Learning from Human Feedback (RLHF) has become a pivotal paradigm for text-to-image generation, its application to image editing remains largely unexplored. A key bottleneck is the lack of a robust general reward model…

Computer Vision and Pattern Recognition · Computer Science 2026-05-21 Hanzhong Guo , Jie Wu , Jie Liu , Yu Gao , Zilyu Ye , Linxiao Yuan , Xionghui Wang , Yizhou Yu , Weilin Huang

Reinforcement Learning with Verifiable Rewards (RLVR) has achieved great success in developing Large Language Models (LLMs) with chain-of-thought rollouts for many tasks such as math and coding. Nevertheless, RLVR struggles with sample…

Machine Learning · Computer Science 2026-05-15 Kai Yan , Alexander G. Schwing , Yu-Xiong Wang

When working to understand usage of a data format, examples of the data format are often more representative than the format's specification. For example, two different applications might use very different JSON representations, or two…

Machine Learning · Computer Science 2021-06-04 Walt Woods

Reinforcement learning from verifiable rewards (RLVR) stimulates the thinking processes of large language models (LLMs), substantially enhancing their reasoning abilities on verifiable tasks. It is often assumed that similar gains should…

Computation and Language · Computer Science 2026-03-24 Kaiyuan Li , Jing-Cheng Pang , Yang Yu

Learning from rewards (i.e., reinforcement learning or RL) and learning to imitate a teacher (i.e., teacher-student learning) are two established approaches for solving sequential decision-making problems. To combine the benefits of these…

Machine Learning · Computer Science 2024-02-21 Idan Shenfeld , Zhang-Wei Hong , Aviv Tamar , Pulkit Agrawal

Reinforcement Learning with Verifiable Rewards (RLVR) is an effective paradigm for improving the reasoning capabilities of large language models. However, existing RLVR methods utilize rollouts in an indiscriminate and short-horizon manner:…

Machine Learning · Computer Science 2026-05-26 Xiaodong Lu , Xiaohan Wang , Jiajun Chai , Guojun Yin , Wei Lin , Zhijun Chen , Yu Luo , Fuzhen Zhuang , Yikun Ban , Deqing Wang

Monitoring the chain-of-thought (CoT) of reasoning models is a promising approach for detecting covert misbehavior (i.e., hidden objectives) in code generation tasks. While large models (GPT-5, Gemini-3-Flash) can serve as effective CoT…

Cryptography and Security · Computer Science 2026-05-14 Nirav Diwan , Han Wang , Berkcan Kapusuzoglu , Ramin Moradi , Supriyo Chakraborty , Giri Iyengar , Sambit Sahu , Huan Zhang , Gang Wang

Reinforcement Learning from Human Feedback (RLHF) remains vulnerable to reward hacking, where models exploit spurious correlations in learned reward models to achieve high scores while violating human intent. Existing mitigations rely on…

Artificial Intelligence · Computer Science 2026-02-03 Mohammad Beigi , Ming Jin , Junshan Zhang , Qifan Wang , Lifu Huang

Chain of thought reasoning has demonstrated remarkable success in large language models, yet its adaptation to vision-language reasoning remains an open challenge with unclear best practices. Existing attempts typically employ reasoning…

Computer Vision and Pattern Recognition · Computer Science 2025-09-24 Honghao Chen , Xingzhou Lou , Xiaokun Feng , Kaiqi Huang , Xinlong Wang

Reinforcement learning with verifiable rewards (RLVR) has become a key technique for enhancing LLMs' reasoning abilities, yet its data inefficiency remains a major bottleneck. To address this critical yet challenging issue, we present a…

Machine Learning · Computer Science 2026-04-28 Shipeng Li , Zhiqin Yang , Shikun Li , Xiaobo Xia , Hengyu Liu , Xinghua Zhang , Gaode Chen , Dong Fang , Ying Tai , Zhe Peng

Outcome-reward reinforcement learning (RL) has proven effective at improving the reasoning capabilities of large language models (LLMs). However, standard RL assigns credit only at the level of the final answer, penalizing entire reasoning…

Machine Learning · Computer Science 2026-01-21 Matthew Y. R. Yang , Hao Bai , Ian Wu , Gene Yang , Amrith Setlur , Aviral Kumar

Group Relative Policy Optimization (GRPO) has been shown to be an effective algorithm when an accurate reward model is available. However, such a highly reliable reward model is not available in many real-world tasks. In this paper, we…

Machine Learning · Computer Science 2026-01-12 Yuki Ichihara , Yuu Jinnai , Tetsuro Morimura , Mitsuki Sakamoto , Ryota Mitsuhashi , Eiji Uchibe