Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.07782 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915723790516224 |
|---|---|
| author | Fang, Wei Glass, James |
| author_facet | Fang, Wei Glass, James |
| contents | LLM agents operating over massive, dynamic tool libraries rely on effective retrieval, yet standard single-shot dense retrievers struggle with complex requests. These failures primarily stem from the disconnect between abstract user goals and technical documentation, and the limited capacity of fixed-size embeddings to model combinatorial tool compositions. To address these challenges, we propose TOOLQP, a lightweight framework that models retrieval as iterative query planning. Instead of single-shot matching, TOOLQP decomposes instructions into sub-tasks and dynamically generates queries to interact with the retriever, effectively bridging the semantic gap by targeting the specific sub-tasks required for composition. We train TOOLQP using synthetic query trajectories followed by optimization via Reinforcement Learning with Verifiable Rewards (RLVR). Experiments demonstrate that TOOLQP achieves state-of-the-art performance, exhibiting superior zero-shot generalization, robustness across diverse retrievers, and significant improvements in downstream agentic execution. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_07782 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Beyond Single-Shot: Multi-step Tool Retrieval via Query Planning Fang, Wei Glass, James Computation and Language Artificial Intelligence Information Retrieval LLM agents operating over massive, dynamic tool libraries rely on effective retrieval, yet standard single-shot dense retrievers struggle with complex requests. These failures primarily stem from the disconnect between abstract user goals and technical documentation, and the limited capacity of fixed-size embeddings to model combinatorial tool compositions. To address these challenges, we propose TOOLQP, a lightweight framework that models retrieval as iterative query planning. Instead of single-shot matching, TOOLQP decomposes instructions into sub-tasks and dynamically generates queries to interact with the retriever, effectively bridging the semantic gap by targeting the specific sub-tasks required for composition. We train TOOLQP using synthetic query trajectories followed by optimization via Reinforcement Learning with Verifiable Rewards (RLVR). Experiments demonstrate that TOOLQP achieves state-of-the-art performance, exhibiting superior zero-shot generalization, robustness across diverse retrievers, and significant improvements in downstream agentic execution. |
| title | Beyond Single-Shot: Multi-step Tool Retrieval via Query Planning |
| topic | Computation and Language Artificial Intelligence Information Retrieval |
| url | https://arxiv.org/abs/2601.07782 |