Beyond ReAct: A Planner-Centric Framework for Complex Tool-Augmented LLM Reasoning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908672833093632 |
|---|---|
| author | Wei, Xiaolong Dong, Yuehu Wang, Xingliang Zhang, Xingyu Zhao, Zhejun Shen, Dongdong Xia, Long Yin, Dawei |
| author_facet | Wei, Xiaolong Dong, Yuehu Wang, Xingliang Zhang, Xingyu Zhao, Zhejun Shen, Dongdong Xia, Long Yin, Dawei |
| contents | Existing tool-augmented large language models (LLMs) encounter significant challenges when processing complex queries. Current frameworks such as ReAct are prone to local optimization traps due to their reliance on incremental decision-making processes. To address these limitations, we propose a novel Planner-centric Plan-Execute paradigm that fundamentally resolves local optimization bottlenecks through architectural innovation. Central to our approach is a novel Planner model that performs global Directed Acyclic Graph (DAG) planning for complex queries, enabling optimized execution beyond conventional tool coordination. We also introduce ComplexTool-Plan, a large-scale benchmark dataset featuring complex queries that demand sophisticated multi-tool composition and coordination capabilities. Additionally, we develop a two-stage training methodology that integrates Supervised Fine-Tuning (SFT) with Group Relative Policy Optimization (GRPO), systematically enhancing the Planner's tool selection accuracy and global planning awareness through structured DAG-based planning. When integrated with a capable executor, our framework achieves state-of-the-art performance on the StableToolBench benchmark for complex user queries, demonstrating superior end-to-end execution capabilities and robust handling of intricate multi-tool workflows. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_10037 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Beyond ReAct: A Planner-Centric Framework for Complex Tool-Augmented LLM Reasoning Wei, Xiaolong Dong, Yuehu Wang, Xingliang Zhang, Xingyu Zhao, Zhejun Shen, Dongdong Xia, Long Yin, Dawei Artificial Intelligence Existing tool-augmented large language models (LLMs) encounter significant challenges when processing complex queries. Current frameworks such as ReAct are prone to local optimization traps due to their reliance on incremental decision-making processes. To address these limitations, we propose a novel Planner-centric Plan-Execute paradigm that fundamentally resolves local optimization bottlenecks through architectural innovation. Central to our approach is a novel Planner model that performs global Directed Acyclic Graph (DAG) planning for complex queries, enabling optimized execution beyond conventional tool coordination. We also introduce ComplexTool-Plan, a large-scale benchmark dataset featuring complex queries that demand sophisticated multi-tool composition and coordination capabilities. Additionally, we develop a two-stage training methodology that integrates Supervised Fine-Tuning (SFT) with Group Relative Policy Optimization (GRPO), systematically enhancing the Planner's tool selection accuracy and global planning awareness through structured DAG-based planning. When integrated with a capable executor, our framework achieves state-of-the-art performance on the StableToolBench benchmark for complex user queries, demonstrating superior end-to-end execution capabilities and robust handling of intricate multi-tool workflows. |
| title | Beyond ReAct: A Planner-Centric Framework for Complex Tool-Augmented LLM Reasoning |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2511.10037 |