ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908780471517184 |
|---|---|
| author | Zhang, Qiang Chen, Boli Zhang, Fanrui Ding, Ruixue Wang, Shihang Wang, Qiuchen Huang, Yinfeng Zhang, Haonan Zhu, Rongxiang Wang, Pengyong Ren, Ailin Li, Xin Xie, Pengjun Liu, Jiawei Guo, Ning Zhou, Jingren Zha, Zheng-Jun |
| author_facet | Zhang, Qiang Chen, Boli Zhang, Fanrui Ding, Ruixue Wang, Shihang Wang, Qiuchen Huang, Yinfeng Zhang, Haonan Zhu, Rongxiang Wang, Pengyong Ren, Ailin Li, Xin Xie, Pengjun Liu, Jiawei Guo, Ning Zhou, Jingren Zha, Zheng-Jun |
| contents | Reinforcement learning has substantially improved the performance of LLM agents on tasks with verifiable outcomes, but it still struggles on open-ended agent tasks with vast solution spaces (e.g., complex travel planning). Due to the absence of objective ground-truth for these tasks, current RL algorithms largely rely on reward models that assign scalar scores to individual responses. We contend that such pointwise scoring suffers from an inherent discrimination collapse: the reward model struggles to distinguish subtle advantages among different trajectories, resulting in scores within a group being compressed into a narrow range. Consequently, the effective reward signal becomes dominated by noise from the reward model, leading to optimization stagnation. To address this, we propose ArenaRL, a reinforcement learning paradigm that shifts from pointwise scalar scoring to intra-group relative ranking. ArenaRL introduces a process-aware pairwise evaluation mechanism, employing multi-level rubrics to assign fine-grained relative scores to trajectories. Additionally, we construct an intra-group adversarial arena and devise a tournament-based ranking scheme to obtain stable advantage signals. Empirical results confirm that the built seeded single-elimination scheme achieves nearly equivalent advantage estimation accuracy to full pairwise comparisons with O(N^2) complexity, while operating with only O(N) complexity, striking an optimal balance between efficiency and precision. Furthermore, to address the lack of full-cycle benchmarks for open-ended agents, we build Open-Travel and Open-DeepResearch, two high-quality benchmarks featuring a comprehensive pipeline covering SFT, RL training, and multi-dimensional evaluation. Extensive experiments show that ArenaRL substantially outperforms standard RL baselines, enabling LLM agents to generate more robust solutions for complex real-world tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_06487 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking Zhang, Qiang Chen, Boli Zhang, Fanrui Ding, Ruixue Wang, Shihang Wang, Qiuchen Huang, Yinfeng Zhang, Haonan Zhu, Rongxiang Wang, Pengyong Ren, Ailin Li, Xin Xie, Pengjun Liu, Jiawei Guo, Ning Zhou, Jingren Zha, Zheng-Jun Machine Learning Artificial Intelligence Reinforcement learning has substantially improved the performance of LLM agents on tasks with verifiable outcomes, but it still struggles on open-ended agent tasks with vast solution spaces (e.g., complex travel planning). Due to the absence of objective ground-truth for these tasks, current RL algorithms largely rely on reward models that assign scalar scores to individual responses. We contend that such pointwise scoring suffers from an inherent discrimination collapse: the reward model struggles to distinguish subtle advantages among different trajectories, resulting in scores within a group being compressed into a narrow range. Consequently, the effective reward signal becomes dominated by noise from the reward model, leading to optimization stagnation. To address this, we propose ArenaRL, a reinforcement learning paradigm that shifts from pointwise scalar scoring to intra-group relative ranking. ArenaRL introduces a process-aware pairwise evaluation mechanism, employing multi-level rubrics to assign fine-grained relative scores to trajectories. Additionally, we construct an intra-group adversarial arena and devise a tournament-based ranking scheme to obtain stable advantage signals. Empirical results confirm that the built seeded single-elimination scheme achieves nearly equivalent advantage estimation accuracy to full pairwise comparisons with O(N^2) complexity, while operating with only O(N) complexity, striking an optimal balance between efficiency and precision. Furthermore, to address the lack of full-cycle benchmarks for open-ended agents, we build Open-Travel and Open-DeepResearch, two high-quality benchmarks featuring a comprehensive pipeline covering SFT, RL training, and multi-dimensional evaluation. Extensive experiments show that ArenaRL substantially outperforms standard RL baselines, enabling LLM agents to generate more robust solutions for complex real-world tasks. |
| title | ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2601.06487 |