Research map / Reinforcement learning / RL from feedback
Group Verification-based Policy Optimization for Interactive Coding Agents
Silong Dai, Changzhi Sun, Haolun Wu, Huanran Zheng, Tao Ji, Junchi Yan, Yuanbin Wu, Dell Zhang, Xiaoling Wang, Xuelong Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use sequence-mean-token-mean instead of token-mean aggregation (ablation) | GVPO · AppWorld Test-C | TGC: 34.6 -> 49.4 | Improved |
| use sequence-mean-token-mean instead of sequence-mean-token-sum aggregation (ablation) | GVPO · AppWorld Test-C | TGC: 42.8 -> 49.4 | Improved |
| use asymmetric clipping instead of symmetric clipping (ablation) | GVPO · AppWorld Test-C | TGC: 40.5 -> 49.4 | Improved |
| do not apply std normalization to advantages (ablation) | GVPO · AppWorld Test-C | TGC: 42.8 -> 49.4 | Improved |
| match micro-batch and group sizes (ablation) | GVPO · AppWorld Test-C | TGC: 43.1 -> 49.4 | Improved |
| use multiplicative shaping instead of additive-only shaping (ablation) | GVPO · AppWorld Test-C | TGC: 39.8 -> 49.4 | Improved |
| add multiplicative shaping on Test-N (ablation) | GVPO · AppWorld Test-N | TGC: 64.9 -> 72.6 | Improved |
| use asymmetric clipping on Test-N (ablation) | GVPO · AppWorld Test-N | TGC: 60.1 -> 72.6 | Improved |
| lower shaping coefficient b from 0.2 to 0.1 (ablation) | GVPO · AppWorld Test-C | TGC: 49.4 -> 53.1 | Mixed |
| raise shaping coefficient b from 0.2 to 0.4 (ablation) | GVPO · AppWorld Test-N | TGC: 72.6 -> 67.9 | Hurt |
| add process-verifiable shaping to outcome-only RL (GVPO vs LOOP) | LOOP · AppWorld Test-C | TGC: 45.7 -> 49.4 | Improved |
| RL fine-tune Qwen2.5-32B with GVPO versus zero-shot prompting | Qwen2.5-32B-Instruct · AppWorld Test-C | TGC: 18.9 -> 49.4 | Improved |
Closest papers by meaning
- Group-in-Group Policy Optimization for LLM Agent Training (NeurIPS 2025 · similarity 0.90)
- MURPHY: Feedback-Aware GRPO with Retrospective Credit Assignment for Multi-Turn Code Generation (NeurIPS 2026 · similarity 0.89)
- Agentic Reinforced Policy Optimization (ICLR 2026 · similarity 0.88)
- Stronger-MAS: Multi-Agent Reinforcement Learning for Collaborative LLMs (ICLR 2026 · similarity 0.88)
- PAIR: Prefix-Aware Internal Reward Model for Multi-Turn Agent Optimization (NeurIPS 2026 · similarity 0.88)
- Group-Graph Policy Optimization for Long-Horizon Agentic Reinforcement Learning (NeurIPS 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
