Tevatron/AgentIR-data
收藏Hugging Face2026-03-11 更新2026-04-05 收录
下载链接：
https://hf-mirror.com/datasets/Tevatron/AgentIR-data
下载链接
链接失效反馈官方服务：
资源简介：
---
task_categories:
- text-retrieval
language:
- en
tags:
- agent
---
This dataset contains the DR-Synth generated [WebShaper](https://huggingface.co/datasets/Alibaba-NLP/WebShaper) data for training [AgentIR-4B](https://huggingface.co/Tevatron/AgentIR-4B).
- **Paper:** [AgentIR: Reasoning-Aware Retrieval for Deep Research Agents](https://huggingface.co/papers/2603.04384)
- **Code:** [https://github.com/texttron/AgentIR](https://github.com/texttron/AgentIR)
- **Project Page:** [https://texttron.github.io/AgentIR/](https://texttron.github.io/AgentIR/)
- **Model:** [AgentIR-4B](https://huggingface.co/Tevatron/AgentIR-4B)
## Dataset Details
Each instance contains:
- `query_id`: `{webshaper_query_id}_turn{i}`, where `webshaper_query_id` is the original id in WebShaper, and `i` is the turn number during agent rollout when constructing the data.
- `query`: the reasoning-concatenated query that the agent issued during rollout
- `positive_passages`: list of positive documents, where each document contains the "docid" in the training corpus, and "text" is the content of the document.
- `negative_passages`: list of negative documents, where each document follows a similar structure as above.
## Citation
```bibtex
@article{chen2026AgentIR,
title={AgentIR: Reasoning-Aware Retrieval for Deep Research Agents},
author={Zijian Chen and Xueguang Ma and Shengyao Zhuang and Jimmy Lin and Akari Asai and Victor Zhong},
year={2026},
journal={arXiv preprint arXiv:2603.04384}
}
```
提供机构：
Tevatron
