paper-with-me

Papers

CM2: Reinforcement Learning with Checklist Rewards for Multi-Turn and Multi-Step Agentic Tool Use

2026-02-12 · Zhen Zhang, Kaiqiang Song, Xun Wang, Yebowen Hu, Weixiang Yan, Chenyang Zhao, Henry Peng Zou, Haoyun Deng, Sathish Reddy Indurthi, Shujian Liu, Simin Ma, Xiaoyang Wang, Xin Eric Wang, Song Wang arxiv

AI agents are increasingly used to solve real-world tasks by reasoning over multi-turn user interactions and invoking external tools. However, applying reinforcement learning to such settings remains difficult: realistic objectives often lack verifiable rewards and instead emphasize open-ended behaviors; moreover, RL for multi-turn, multi-step agentic tool use is still underexplored; and building and maintaining executable tool environments is costly, limiting scale and coverage. We propose CM2, an RL framework that replaces verifiable outcome rewards with checklist rewards. CM2 decomposes each turn's intended behavior into fine-grained binary criteria with explicit evidence grounding and structured metadata, turning open-ended judging into more stable classification-style decisions. To balance stability and informativeness, our method adopts a strategy of sparse reward assignment but dense evaluation criteria. Training is performed in a scalable LLM-simulated tool environment, avoiding heavy engineering for large tool sets. Experiments show that CM2 consistently improves over supervised fine-tuning. Starting from an 8B Base model and training on an 8k-example RL dataset, CM2 improves over the SFT counterpart by 8 points on tau^-Bench, by 10 points on BFCL-V4, and by 12 points on ToolSandbox. The results match or even outperform similarly sized open-source baselines, including the judging model. CM2 thus provides a scalable recipe for optimizing multi-turn, multi-step tool-using agents without relying on verifiable rewards. Code provided by the open-source community: https://github.com/namezhenzhang/CM2-RLCR-Tool-Agent.

📄 PDF Abstract BibTeX arXiv:2602.12268

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Soft-SVeRL: Self-Verified Reinforcement Learning with Soft Rewards

2026-05-27 · Saurabh Dash, Pierre Clavier, John Dang, Matthias Galle 외 arxiv

Reinforcement Learning from Verifiable Rewards (RLVR) has improved language models in domains such as mathematics and code, where correctness can be checked automatically. However, many important tasks are only partially…

Reinforcement Learning

Checklists Are Better Than Reward Models For Aligning Language Models

2025-07-24 · Vijay Viswanathan, Yanchao Sun, Shuang Ma, Xiang Kong 외 arxiv

Language models must be adapted to understand and follow user instructions. Reinforcement learning is widely used to facilitate this -- typically using fixed criteria such as "helpfulness" and "harmfulness". In our work,…

Reinforcement LearningInstruction Following

From Reasoning to Agentic: Credit Assignment in Reinforcement Learning for Large Language Models

2026-04-10 · Chenchen Zhang arxiv

Reinforcement learning (RL) for large language models (LLMs) increasingly relies on sparse, outcome-level rewards -- yet determining which actions within a long trajectory caused the outcome remains difficult. This credi…

Reinforcement Learning

Rewarding Better Thinking for LLM Preference Alignment

2026-07-22 · Xubo Liu, Wenya Guo, Ruxue Yan, Xinying Qian 외 arxiv

LLM preference alignment aims to optimize models toward human preferences across diverse user instructions. Reinforcement learning has become a major post-training approach for this goal, but existing proxy rewards are o…

Reinforcement Learning

EvoIdeator: Evolving Scientific Ideas through Checklist-Grounded Reinforcement Learning

2026-03-23 · Andreas Sauter, Yuyue Zhao, Jacopo Urbani, Wenxiang Hu 외 arxiv

Scientific idea generation is a cornerstone of autonomous knowledge discovery, yet the iterative evolution required to transform initial concepts into high-quality research proposals remains a formidable challenge for La…

Reinforcement Learning