Reward Models Identify Consistency, Not Causality

• URL: http://arxiv.org/abs/2502.14619v1
• Date: Thu, 20 Feb 2025 14:57:14 GMT
• Title: Reward Models Identify Consistency, Not Causality
• Authors: Yuhui Xu, Hanze Dong, Lei Wang, Caiming Xiong, Junnan Li,
• Abstract summary: State-of-the-art reward models prioritize structural consistency over causal correctness. Removing the problem statement has minimal impact on reward scores. altering numerical values or disrupting the reasoning flow significantly affects RM outputs.
• Score: 54.987590763737145
• License:
• Abstract: Reward models (RMs) play a crucial role in aligning large language models (LLMs) with human preferences and enhancing reasoning quality. Traditionally, RMs are trained to rank candidate outputs based on their correctness and coherence. However, in this work, we present several surprising findings that challenge common assumptions about RM behavior. Our analysis reveals that state-of-the-art reward models prioritize structural consistency over causal correctness. Specifically, removing the problem statement has minimal impact on reward scores, whereas altering numerical values or disrupting the reasoning flow significantly affects RM outputs. Furthermore, RMs exhibit a strong dependence on complete reasoning trajectories truncated or incomplete steps lead to significant variations in reward assignments, indicating that RMs primarily rely on learned reasoning patterns rather than explicit problem comprehension. These findings hold across multiple architectures, datasets, and tasks, leading to three key insights: (1) RMs primarily assess coherence rather than true reasoning quality; (2) The role of explicit problem comprehension in reward assignment is overstated; (3) Current RMs may be more effective at ranking responses than verifying logical validity. Our results suggest a fundamental limitation in existing reward modeling approaches, emphasizing the need for a shift toward causality-aware reward models that go beyond consistency-driven evaluation.

Related papers

• Retrieval-Augmented Process Reward Model for Generalizable Mathematical Reasoning [32.850036320802474]
  We introduce Retrieval-Augmented Process Reward Model (RetrievalPRM), a novel framework designed to tackle OOD issues. By utilizing a two-stage retrieval-enhanced mechanism, RetrievalPRM retrieves semantically similar questions and steps as a warmup. Our experiments demonstrate that RetrievalPRM outperforms existing baselines across multiple real-world datasets.
  arXiv  Detail & Related papers  (2025-02-20T08:40:09Z)
• ReARTeR: Retrieval-Augmented Reasoning with Trustworthy Process Rewarding [25.329712997545794]
  We propose Retrieval-Augmented Reasoning through Trustworthy Process Rewarding (ReARTeR) ReARTeR enhances RAG systems' reasoning capabilities through post-training and test-time scaling. Experimental results on multi-step reasoning benchmarks demonstrate significant improvements.
  arXiv  Detail & Related papers  (2025-01-14T05:56:26Z)
• Self-Generated Critiques Boost Reward Modeling for Language Models [57.60881438647227]
  Critic-RM is a framework that improves reward models using self-generated critiques without extra supervision. Experiments show that Critic-RM improves reward modeling accuracy by 3.7%-7.3% compared to standard reward models and LLM judges.
  arXiv  Detail & Related papers  (2024-11-25T18:28:26Z)
• RMB: Comprehensively Benchmarking Reward Models in LLM Alignment [44.84304822376291]
  Reward models (RMs) guide the alignment of large language models (LLMs) We propose RMB, a comprehensive RM benchmark that covers over 49 real-world scenarios. Based on our benchmark, we conduct extensive analysis on the state-of-the-art RMs.
  arXiv  Detail & Related papers  (2024-10-13T16:06:54Z)
• Boosting the Power of Small Multimodal Reasoning Models to Match Larger Models with Self-Consistency Training [49.3242278912771]
  Multimodal reasoning is a challenging task that requires models to reason across multiple modalities to answer questions. Existing approaches have made progress by incorporating language and visual modalities into a two-stage reasoning framework. We propose MC-CoT, a self-consistency training strategy that generates multiple rationales and answers, subsequently selecting the most accurate through a voting process.
  arXiv  Detail & Related papers  (2023-11-23T17:09:48Z)
• Confronting Reward Model Overoptimization with Constrained RLHF [114.71591361764547]
  We show that correlation between component RMs has a significant effect on the locations of these points. Our method addresses the problem of weighting component RMs by learning dynamic weights, naturally expressed by Lagrange multipliers.
  arXiv  Detail & Related papers  (2023-10-06T16:59:17Z)
• The Trickle-down Impact of Reward (In-)consistency on RLHF [71.37987812944971]
  We show that reward inconsistency exhibits a trickle-down effect on the downstream Reinforcement Learning from Human Feedback process. We propose Contrast Instructions -- a benchmarking strategy for the consistency of RM. We show that RLHF models trained with a more consistent RM yield more useful responses.
  arXiv  Detail & Related papers  (2023-09-28T04:05:13Z)
• Context De-confounded Emotion Recognition [12.037240778629346]
  Context-Aware Emotion Recognition (CAER) aims to perceive the emotional states of the target person with contextual information. A long-overlooked issue is that a context bias in existing datasets leads to a significantly unbalanced distribution of emotional states. This paper provides a causality-based perspective to disentangle the models from the impact of such bias, and formulate the causalities among variables in the CAER task.
  arXiv  Detail & Related papers  (2023-03-21T15:12:20Z)
• Learning Causal Semantic Representation for Out-of-Distribution Prediction [125.38836464226092]
  We propose a Causal Semantic Generative model (CSG) based on a causal reasoning so that the two factors are modeled separately. We show that CSG can identify the semantic factor by fitting training data, and this semantic-identification guarantees the boundedness of OOD generalization error.
  arXiv  Detail & Related papers  (2020-11-03T13:16:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.