paper-with-me

Papers

Beyond Reward Hacking: Causal Rewards for Large Language Model Alignment

2025-01-16 · Chaoqi Wang, Zhuokai Zhao, Yibo Jiang, Zhaorun Chen, Chen Zhu, Yuxin Chen, Jiayi Liu, Lizhu Zhang, Xiangjun Fan, Hao Ma, Sinong Wang

Recent advances in large language models (LLMs) have demonstrated significant progress in performing complex tasks. While Reinforcement Learning from Human Feedback (RLHF) has been effective in aligning LLMs with human preferences, it is susceptible to spurious correlations in reward modeling. Consequently, it often introduces biases-such as length bias, sycophancy, conceptual bias, and discrimination that hinder the model's ability to capture true causal relationships. To address this, we propose a novel causal reward modeling approach that integrates causal inference to mitigate these spurious correlations. Our method enforces counterfactual invariance, ensuring reward predictions remain consistent when irrelevant variables are altered. Through experiments on both synthetic and real-world datasets, we show that our approach mitigates various types of spurious correlations effectively, resulting in more reliable and fair alignment of LLMs with human preferences. As a drop-in enhancement to the existing RLHF workflow, our causal reward modeling provides a practical way to improve the trustworthiness and fairness of LLM finetuning.

📄 PDF Abstract BibTeX arXiv:2501.09620

Code (1)

tatsu-lab/alpaca_farm 공식 구현 pytorch

Tasks

Causal InferencecounterfactualFairnessLanguage ModelingLanguage ModellingLarge Language Model

Methods 이 논문이 사용한 방법론

Causal inference Causal inference is the process of drawing a conclusion about a causal connection based on the conditions of the occurrence of an effect. The main difference between causal…

Similar Papers 제목 키워드 기반

Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR

2026-03-07 · Muhammad Khalifa, Zohaib Khan, Omer Tafveez, Hao Peng 외 arxiv

Reward hacking is a form of misalignment in which models overoptimize proxy rewards without genuinely solving the underlying task. Precisely measuring reward hacking occurrence remains challenging because true task rewar…

Reinforcement LearningMathematical Reasoning

Multimodal Reward Hacking in Reinforcement Learning

2026-07-10 · Jiayu Yao, Yiwei Wang, Anmeng Zhang, Zhe Sun 외 arxiv

Reinforcement learning (RL) is increasingly used to align multimodal large language models (MLLMs), but higher rewards do not always imply better task performance. This risk is amplified when visual evidence is evaluated…

Reinforcement Learning

Causal Confusion and Reward Misidentification in Preference-Based Reward Learning

2022-04-13 · Jeremy Tien, Jerry Zhi-Yang He, Zackory Erickson, Anca D. Dragan 외

Learning policies via preference-based reward learning is an increasingly popular method for customizing agent behavior, but has been shown anecdotally to be prone to spurious correlations and reward hacking behaviors. W…

Imitation Learning

Proof-of-Use: Mitigating Tool-Call Hacking in Deep Research Agents

2025-10-13 · SHengjie Ma, Chenlong Deng, Jiaxin Mao, Jiadeng Huang 외 arxiv

While reinforcement learning (RL) enhances their ability to plan and reason across retrieval steps, we identify a critical failure mode in this setting: Tool-Call Hacking. Unlike execution-based tools (e.g., code or math…

Reinforcement Learning

Linking Process to Outcome: Conditional Reward Modeling for LLM Reasoning

2025-09-30 · Zheng Zhang, Ziwei Shan, Kaitao Song, Yexin Li 외 arxiv

Process Reward Models (PRMs) have emerged as a promising approach to enhance the reasoning capabilities of large language models (LLMs) by guiding their step-by-step reasoning toward a final answer. However, existing PRM…

Reinforcement Learning