Cover Image

Harder Is Better: Boosting Mathematical Reasoning via Difficulty-Aware GRPO and Multi-Aspect Question Reformulation

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Dai, Yanqi, Ji, Yuxiang, Zhang, Xiao, Wang, Yong, Chu, Xiangxiang, Lu, Zhiwu
Format:	Preprint
Published:	2026
Subjects:	Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Adaptive Task Balancing for Visual Instruction Tuning via Inter-Task Contribution and Intra-Task Difficulty
by: Dai, Yanqi, et al.
Published: (2024)

GanitLLM: Difficulty-Aware Bengali Mathematical Reasoning through Curriculum-GRPO
by: Dipta, Shubhashis Roy, et al.
Published: (2026)

HS-STaR: Hierarchical Sampling for Self-Taught Reasoners via Difficulty Estimation and Budget Reallocation
by: Xiong, Feng, et al.
Published: (2025)

D$^2$Evo: Dual Difficulty-Aware Self-Evolution for Data-Efficient Reinforcement Learning
by: Zhang, Ru, et al.
Published: (2026)

GRPO-LEAD: A Difficulty-Aware Reinforcement Learning Approach for Concise Mathematical Reasoning in Language Models
by: Zhang, Jixiao, et al.
Published: (2025)

NLP Methods May Actually Be Better Than Professors at Estimating Question Difficulty
by: Zotos, Leonidas, et al.
Published: (2025)

Multi-Task GRPO: Reliable LLM Reasoning Across Tasks
by: Ramesh, Shyam Sundhar, et al.
Published: (2026)

DART-Math: Difficulty-Aware Rejection Tuning for Mathematical Problem-Solving
by: Tong, Yuxuan, et al.
Published: (2024)

Benchmarking LLMs' Mathematical Reasoning with Unseen Random Variables Questions
by: Hong, Zijin, et al.
Published: (2025)

AdaCtrl: Towards Adaptive and Controllable Reasoning via Difficulty-Aware Budgeting
by: Huang, Shijue, et al.
Published: (2025)

SkillClaw: Let Skills Evolve Collectively with Agentic Evolver
by: Ma, Ziyu, et al.
Published: (2026)

Conversational Question Answering with Reformulations over Knowledge Graph
by: Liu, Lihui, et al.
Published: (2023)

I Could've Asked That: Reformulating Unanswerable Questions
by: Zhao, Wenting, et al.
Published: (2024)

Critique-GRPO: Advancing LLM Reasoning with Natural Language and Numerical Feedback
by: Zhang, Xiaoying, et al.
Published: (2025)

GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning
by: Chen, Yi, et al.
Published: (2025)

Question Difficulty Ranking for Multiple-Choice Reading Comprehension
by: Raina, Vatsal, et al.
Published: (2024)

DRA-GRPO: Your GRPO Needs to Know Diverse Reasoning Paths for Mathematical Reasoning
by: Chen, Xiwen, et al.
Published: (2025)

MathGenie: Generating Synthetic Data with Question Back-translation for Enhancing Mathematical Reasoning of LLMs
by: Lu, Zimu, et al.
Published: (2024)

Fast-Slow Thinking GRPO for Large Vision-Language Model Reasoning
by: Xiao, Wenyi, et al.
Published: (2025)

The LLM Already Knows: Estimating LLM-Perceived Question Difficulty via Hidden Representations
by: Zhu, Yubo, et al.
Published: (2025)

Preference Alignment for Diffusion Model via Explicit Denoised Distribution Estimation
by: Shi, Dingyuan, et al.
Published: (2024)

ExGRPO: Learning to Reason from Experience
by: Zhan, Runzhe, et al.
Published: (2025)

Robust Training for Conversational Question Answering Models with Reinforced Reformulation Generation
by: Kaiser, Magdalena, et al.
Published: (2023)

BoostTaxo: Zero-Shot Taxonomy Induction via Boosting-Style Agentic Reasoning and Constraint-Aware Calibration
by: Ling, Yancheng, et al.
Published: (2026)

Temporal-Aware Heterogeneous Graph Reasoning with Multi-View Fusion for Temporal Question Answering
by: Wen, Wuzhenghong, et al.
Published: (2026)

RankLLM: Weighted Ranking of LLMs by Quantifying Question Difficulty
by: Zhang, Ziqian, et al.
Published: (2026)

DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning
by: Shao, Zhihong, et al.
Published: (2025)

Mitigating Selection Bias in Large Language Models via Permutation-Aware GRPO
by: Zheng, Jinquan, et al.
Published: (2026)

StoryCoder: Narrative Reformulation for Structured Reasoning in LLM Code Generation
by: Jang, Geonhui, et al.
Published: (2026)

MMR-GRPO: Accelerating GRPO-Style Training through Diversity-Aware Reward Reweighting
by: Wei, Kangda, et al.
Published: (2026)

The Harder The Better: Maintaining Supervised Fine-tuning Generalization with Less but Harder Data
by: Shang, Zhaoyang, et al.
Published: (2025)

Towards Better Generalization in Open-Domain Question Answering by Mitigating Context Memorization
by: Zhang, Zixuan, et al.
Published: (2024)

From Reasoning to Code: GRPO Optimization for Underrepresented Languages
by: Pennino, Federico, et al.
Published: (2025)

Thinking with Map: Reinforced Parallel Map-Augmented Agent for Geolocalization
by: Ji, Yuxiang, et al.
Published: (2026)

Bridging Writing Manner Gap in Visual Instruction Tuning by Creating LLM-aligned Instructions
by: Jing, Dong, et al.
Published: (2025)

The Zero-Step Thinking: An Empirical Study of Mode Selection as Harder Early Exit in Reasoning Models
by: Tan, Yuqiao, et al.
Published: (2025)

Omne-R1: Learning to Reason with Memory for Multi-hop Question Answering
by: Liu, Boyuan, et al.
Published: (2025)

What Defines Good Reasoning in LLMs? Dissecting Reasoning Steps with Multi-Aspect Evaluation
by: Do, Heejin, et al.
Published: (2025)

BeamAggR: Beam Aggregation Reasoning over Multi-source Knowledge for Multi-hop Question Answering
by: Chu, Zheng, et al.
Published: (2024)

Outdated Issue Aware Decoding for Reasoning Questions on Edited Knowledge
by: Sun, Zengkui, et al.
Published: (2024)