Copertina

Stop Rewarding Hallucinated Steps: Faithfulness-Aware Step-Level Reinforcement Learning for Small Reasoning Models

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Nie, Shuo, Deng, Hexuan, Wang, Chao, Fang, Ruiyu, Liu, Xuebo, Song, Shuangyong, Li, Yu, Zhang, Min, Li, Xuelong
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

REA-RL: Reflection-Aware Online Reinforcement Learning for Efficient Reasoning
di: Deng, Hexuan, et al.
Pubblicazione: (2025)

FaithRL: Learning to Reason Faithfully through Step-Level Faithfulness Maximization
di: Gui, Runquan, et al.
Pubblicazione: (2026)

RouterKGQA: Specialized--General Model Routing for Constraint-Aware Knowledge Graph Question Answering
di: Yuan, Bo, et al.
Pubblicazione: (2026)

MR-UIE: Multi-Perspective Reasoning with Reinforcement Learning for Universal Information Extraction
di: Li, Zhongqiu, et al.
Pubblicazione: (2025)

Rewarding What Matters: Step-by-Step Reinforcement Learning for Task-Oriented Dialogue
di: Du, Huifang, et al.
Pubblicazione: (2024)

Big-Stop Semantics: Small-Step Semantics in a Big-Step Judgment
di: Kahn, David M, et al.
Pubblicazione: (2025)

GroundedPRM: Tree-Guided and Fidelity-Aware Process Reward Modeling for Step-Level Reasoning
di: Zhang, Yao, et al.
Pubblicazione: (2025)

Dynamic Sampling that Adapts: Self-Aware Iterative Data Persistent Optimization for Mathematical Reasoning
di: Rao, Jun, et al.
Pubblicazione: (2025)

Agentic Reinforcement Learning with Implicit Step Rewards
di: Liu, Xiaoqian, et al.
Pubblicazione: (2025)

Introducing Visual Scenes and Reasoning: A More Realistic Benchmark for Spoken Language Understanding
di: Wu, Di, et al.
Pubblicazione: (2025)

StepPO: Step-Aligned Policy Optimization for Agentic Reinforcement Learning
di: Wang, Daoyu, et al.
Pubblicazione: (2026)

Emotional Support with LLM-based Empathetic Dialogue Generation
di: Wang, Shiquan, et al.
Pubblicazione: (2025)

D-QRELO: Training- and Data-Free Delta Compression for Large Language Models via Quantization and Residual Low-Rank Approximation
di: Li, Junlin, et al.
Pubblicazione: (2026)

UniARM: Towards a Unified Autoregressive Reward Model for Multi-Objective Test-Time Alignment
di: Xie, Hongyan, et al.
Pubblicazione: (2026)

Prompt-Level Reward Specifications for Open-Ended Post-Training
di: Weng, Zijun, et al.
Pubblicazione: (2026)

Let's Reward Step-by-Step: Step-Aware Contrastive Alignment for Vision-Language Navigation in Continuous Environments
di: Li, Haoyuan, et al.
Pubblicazione: (2026)

RB-SQL: A Retrieval-based LLM Framework for Text-to-SQL
di: Wu, Zhenhe, et al.
Pubblicazione: (2024)

R3-RAG: Learning Step-by-Step Reasoning and Retrieval for LLMs via Reinforcement Learning
di: Li, Yuan, et al.
Pubblicazione: (2025)

Hierarchical Reinforcement Learning with Augmented Step-Level Transitions for LLM Agents
di: Zhen, Shuai, et al.
Pubblicazione: (2026)

NewTerm: Benchmarking Real-Time New Terms for Large Language Models with Annual Updates
di: Deng, Hexuan, et al.
Pubblicazione: (2024)

DRPruning: Efficient Large Language Model Pruning through Distributionally Robust Optimization
di: Deng, Hexuan, et al.
Pubblicazione: (2024)

Truncated Step-Level Sampling with Process Rewards for Retrieval-Augmented Reasoning
di: Samarinas, Chris, et al.
Pubblicazione: (2026)

Multi-Step Likelihood-Ratio Correction for Reinforcement Learning with Verifiable Rewards
di: Yoon, Deokgyu, et al.
Pubblicazione: (2026)

Measuring Chain of Thought Faithfulness by Unlearning Reasoning Steps
di: Tutek, Martin, et al.
Pubblicazione: (2025)

Dynamic Knowledge Fusion for Multi-Domain Dialogue State Tracking
di: Su, Haoxiang, et al.
Pubblicazione: (2026)

AQuilt: Weaving Logic and Self-Inspection into Low-Cost, High-Relevance Data Synthesis for Specialist LLMs
di: Ke, Xiaopeng, et al.
Pubblicazione: (2025)

Awakening Dormant Experts:Counterfactual Routing to Mitigate MoE Hallucinations
di: Hu, Wentao, et al.
Pubblicazione: (2026)

Step Guided Reasoning: Improving Mathematical Reasoning using Guidance Generation and Step Reasoning
di: Cao, Lang, et al.
Pubblicazione: (2024)

StepTool: Enhancing Multi-Step Tool Usage in LLMs via Step-Grained Reinforcement Learning
di: Yu, Yuanqing, et al.
Pubblicazione: (2024)

Full-Step-DPO: Self-Supervised Preference Optimization with Step-wise Rewards for Mathematical Reasoning
di: Xu, Huimin, et al.
Pubblicazione: (2025)

IG-Search: Step-Level Information Gain Rewards for Search-Augmented Reasoning
di: Liang, Zihan, et al.
Pubblicazione: (2026)

Diffusion Model as a Noise-Aware Latent Reward Model for Step-Level Preference Optimization
di: Zhang, Tao, et al.
Pubblicazione: (2025)

StepOPSD: Step-Aware Online Preference Distillation for Agent Reinforcement Learning
di: Zhang, Yanfei, et al.
Pubblicazione: (2026)

What Are Step-Level Reward Models Rewarding? Counterintuitive Findings from MCTS-Boosted Mathematical Reasoning
di: Ma, Yiran, et al.
Pubblicazione: (2024)

Step-wise Rubric Rewards for LLM Reasoning
di: Xie, Weichu, et al.
Pubblicazione: (2026)

Supervised Reinforcement Learning: From Expert Trajectories to Step-wise Reasoning
di: Deng, Yihe, et al.
Pubblicazione: (2025)

Watch Every Step! LLM Agent Learning via Iterative Step-Level Process Refinement
di: Xiong, Weimin, et al.
Pubblicazione: (2024)

TCNN: Triple Convolutional Neural Network Models for Retrieval-based Question Answering System in E-commerce
di: Song, Shuangyong, et al.
Pubblicazione: (2020)

TRACES: Tagging Reasoning Steps for Adaptive Cost-Efficient Early-Stopping
di: Belkhiter, Yannis, et al.
Pubblicazione: (2026)

CM2: Reinforcement Learning with Checklist Rewards for Multi-Turn and Multi-Step Agentic Tool Use
di: Zhang, Zhen, et al.
Pubblicazione: (2026)