A Step Harder, Not a Step Too Far: The Benefits and Limits of Curriculum Difficulty Progression in GRPO

arXiv:2610.13669 · cs.AI · Submitted 2026-10-09 · Read on arXiv

cs.AI

Submitted: 2026-10-09

Updated: 2026-10-09

Comments: 30 pages, 8 figures, 19 tables

Code: https://github.com/NHLling/GRPO-B

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• GRPO — 1× in this paper · explained in 6 episodes

Sources

• Training Verifiers to Solve Math Word Problems
• Prompt Curriculum Learning for Efficient LLM Post-Training
• LoRA: Low-Rank Adaptation of Large Language Models
• No Prompt Left Behind: Exploiting Zero-Variance Prompts in LLM Reinforcement Learning via Entropy-Guided Advantage Shaping
• LIMR: Less is More for RL Scaling
• Understanding R1-Zero-Like Training: A Critical Perspective
• Dynamics-Predictive Sampling for Active RL Finetuning of Large Reasoning Models
• Curriculum Reinforcement Learning from Easy to Hard Tasks Improves LLM Reasoning
• Can Prompt Difficulty be Online Predicted for Accelerating RL Finetuning of Reasoning Models?
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• BOTS: A Unified Framework for Bayesian Online Task Selection in LLM Reinforcement Finetuning
• Efficient Reinforcement Finetuning via Adaptive Curriculum Learning
• Towards High Data Efficiency in Reinforcement Learning with Verifiable Reward
• Scheduling Your LLM Reinforcement Learning with Reasoning Trees
• Qwen2.5 Technical Report
• DAPO: An Open-Source LLM Reinforcement Learning System at Scale
• CurES: From Gradient Analysis to Efficient Curriculum Learning for Reasoning LLMs
• UFO-RL: Uncertainty-Focused Optimization for Efficient Reinforcement Learning Data Selection
• Group Sequence Policy Optimization
• Efficient RLVR Training via Weighted Mutual Information Data Selection

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• Agents' Last Exam
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems