Learning from All-Failed Groups via First-Visit Coverage in Long-Horizon Agentic Tasks

arXiv:2607.22724 · cs.LG, cs.AI · Submitted 2026-07-22 · Read on arXiv

cs.LG, cs.AI

Submitted: 2026-07-22

Updated: 2026-10-09

Code: https://github.com/KeibingYang/ProGPO

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

Terminology

• Agentic Task — 1× in this paper · explained in 1 episode
• Agentic Tasks — 1× in this paper · explained in 1 episode
• Benchmark — 1× in this paper · explained in 2 episodes
• Large Language Model (LLM) — 1× in this paper · explained in 3 episodes

Sources

• GPT-4 Technical Report
• XRPO: Pushing the limits of GRPO with Targeted Exploration and Exploitation
• Beyond Trajectory-Level Attribution: Graph-Based Credit Assignment for Agentic Reinforcement Learning
• Hierarchy-of-Groups Policy Optimization for Long-Horizon Agentic Tasks
• No Prompt Left Behind: Exploiting Zero-Variance Prompts in LLM Reinforcement Learning via Entropy-Guided Advantage Shaping
• Agentic Reinforcement Learning with Implicit Step Rewards
• Understanding R1-Zero-Like Training: A Critical Perspective
• TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization
• NGRPO: Negative-enhanced Group Relative Policy Optimization
• ToolRL: Reward is All Tool Learning Needs
• RIDE: Rewarding Impact-Driven Exploration for Procedurally-Generated Environments
• Proximal Policy Optimization Algorithms
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Gemini: A Family of Highly Capable Multimodal Models
• Kimi k1.5: Scaling Reinforcement Learning with LLMs
• Information Gain-based Policy Optimization: A Simple and Effective Approach for Multi-Turn Search Agents
• SPA-RL: Reinforcing LLM Agents via Stepwise Progress Attribution
• SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
• Qwen3 Technical Report
• ReAct: Synergizing Reasoning and Acting in Language Models

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks