dorsal/arxiv
View SchemaUnlocking Implicit Experience: Synthesizing Tool-Use Trajectories from Text
| Authors | Zhihao Xu, Rumei Li, Jiahuan Li, Rongxiang Weng, Jingang Wang, Xunliang Cai, Xiting Wang |
|---|---|
| Categories | |
| ArXiv ID | 2601.10355vv1 |
| URL | https://arxiv.org/abs/2601.10355 |
| License | http://creativecommons.org/licenses/by/4.0/ |
Abstract
Enabling Large Language Models (LLMs) to effectively utilize tools in multi-turn interactions is essential for building capable autonomous agents. However, acquiring diverse and realistic multi-turn tool-use data remains a significant challenge. In this work, we propose a novel text-based paradigm. We observe that textual corpora naturally contain rich, multi-step problem-solving experiences, which can serve as an untapped, scalable, and authentic data source for multi-turn tool-use tasks. Based on this insight, we introduce GEM, a data synthesis pipeline that enables the generation and extraction of multi-turn tool-use trajectories from text corpora through a four-stage process: relevance filtering, workflow & tool extraction, trajectory grounding, and complexity refinement. To reduce the computational cost, we further train a specialized Trajectory Synthesizer via supervised fine-tuning. This model distills the complex generation pipeline into an efficient, end-to-end trajectory generator. Experiments demonstrate that our GEM-32B achieve a 16.5% improvement on the BFCL V3 Multi-turn benchmark. Our models partially surpass the performance of models trained on {\tau} - bench (Airline and Retail) in-domain data, highlighting the superior generalization capability derived from our text-based synthesis paradigm. Notably, our Trajectory Synthesizer matches the quality of the full pipeline while significantly reducing inference latency and costs.
{
"annotation_id": "f7f273fa-3790-49ae-a863-ffdc0e41008d",
"date_created": "2026-02-17T05:53:23.983000Z",
"date_modified": "2026-02-17T05:53:23.983000Z",
"file_hash": "949d8a6f7d44a314ba0bef0eac8b2a1b7b0cf144d617943417e8bb04331629a1",
"private": false,
"record": {
"abstract": "Enabling Large Language Models (LLMs) to effectively utilize tools in multi-turn interactions is essential for building capable autonomous agents. However, acquiring diverse and realistic multi-turn tool-use data remains a significant challenge. In this work, we propose a novel text-based paradigm. We observe that textual corpora naturally contain rich, multi-step problem-solving experiences, which can serve as an untapped, scalable, and authentic data source for multi-turn tool-use tasks. Based on this insight, we introduce GEM, a data synthesis pipeline that enables the generation and extraction of multi-turn tool-use trajectories from text corpora through a four-stage process: relevance filtering, workflow \u0026 tool extraction, trajectory grounding, and complexity refinement. To reduce the computational cost, we further train a specialized Trajectory Synthesizer via supervised fine-tuning. This model distills the complex generation pipeline into an efficient, end-to-end trajectory generator. Experiments demonstrate that our GEM-32B achieve a 16.5% improvement on the BFCL V3 Multi-turn benchmark. Our models partially surpass the performance of models trained on {\\tau} - bench (Airline and Retail) in-domain data, highlighting the superior generalization capability derived from our text-based synthesis paradigm. Notably, our Trajectory Synthesizer matches the quality of the full pipeline while significantly reducing inference latency and costs.",
"arxiv_id": "2601.10355",
"authors": [
"Zhihao Xu",
"Rumei Li",
"Jiahuan Li",
"Rongxiang Weng",
"Jingang Wang",
"Xunliang Cai",
"Xiting Wang"
],
"categories": [
"cs.CL"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"title": "Unlocking Implicit Experience: Synthesizing Tool-Use Trajectories from Text",
"url": "https://arxiv.org/abs/2601.10355",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "8b793648-76a9-438f-abdc-6d2052b18be8",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}