Xu Guo 郭旭

I am a third-year Ph.D. student at Fudan University, jointly trained at Shanghai Innovation Institute (SII). I am advised by Qipeng Guo and Xipeng Qiu.

Since 2024, I have been a research intern at Shanghai AI Laboratory, contributing to training data for the InternLM series, including knowledge-focused data, instruction following, and agent trajectories.

My research focuses on training data for language models and AI agents. I study data synthesis for pre-training and post-training. My current interests include leveraging user–agent interaction traces for agent training and mid-training for vision-language models (VLMs).

News

All updates

Selected publications

All publications

  1. Synthetic Pre-Pre-Training Improves Language Model Robustness to Noisy Pre-Training Data

    Xu Guo, Runyu Peng, Jian Tong, Yunhua Zhou, Haijun Lv, Zhihui Lu, Qipeng Guo

    NeurIPS 2026

    arXiv
    BibTeX 
    @inproceedings{guo2026synthetic,
  title = {Synthetic Pre-Pre-Training Improves Language Model Robustness to Noisy Pre-Training Data},
  author = {Guo, Xu and Peng, Runyu and Tong, Jian and Zhou, Yunhua and Lv, Haijun and Lu, Zhihui and Guo, Qipeng},
  booktitle = {Advances in Neural Information Processing Systems},
  year = {2026},
}

  2. IFDECORATOR: Wrapping Instruction Following Reinforcement Learning with Verifiable Rewards

    Xu Guo, Tianyi Liang, Tong Jian, Xiaogui Yang, Ling-I Wu, Chenhui Li, Zhihui Lu, Qipeng Guo, Kai Chen

    NeurIPS 2026

    arXiv
    BibTeX 
    @inproceedings{guo2025ifdecorator,
  title = {{IFDECORATOR}: Wrapping Instruction Following Reinforcement Learning with Verifiable Rewards},
  author = {Guo, Xu and Liang, Tianyi and Jian, Tong and Yang, Xiaogui and Wu, Ling-I and Li, Chenhui and Lu, Zhihui and Guo, Qipeng and Chen, Kai},
  booktitle = {Advances in Neural Information Processing Systems},
  year = {2026},
}

  3. When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling

    Xu Guo, Jian Tong, Zhihui Lu, Qipeng Guo

    EMNLP Findings 2026

    arXiv
    BibTeX 
    @inproceedings{guo2026when,
  title = {When Does Generating More Help? Disentangling {Fixed-Source Synthesis} from {Source Expansion} in {Synthetic Data Scaling}},
  author = {Guo, Xu and Tong, Jian and Lu, Zhihui and Guo, Qipeng},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2026},
  year = {2026},
}

  4. Rethinking Multiple-Choice Questions for RLVR: Unlocking Potential via Distractor Design

    Xu Guo, Qiming Ge, Jian Tong, Kedi Chen, Jin Zhang, Xiaogui Yang, Xuan Gao, Haijun Lv, Zhihui Lu, Yicheng Zou, Qipeng Guo

    ACL Findings 2026

    arXiv Paper
    BibTeX 
    @inproceedings{guo2026rethinking,
  title = {Rethinking Multiple-Choice Questions for {RLVR}: Unlocking Potential via Distractor Design},
  author = {Guo, Xu and Ge, Qiming and Tong, Jian and Chen, Kedi and Zhang, Jin and Yang, Xiaogui and Gao, Xuan and Lv, Haijun and Lu, Zhihui and Zou, Yicheng and Guo, Qipeng},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
}

  5. Towards transferable adversarial attacks on vision transformers for image classification

    Xu Guo, Peng Chen, Zhihui Lu, Hongfeng Chai, Xin Du, Xudong Wu

    JSA 2024 (CCF-B)

    Paper DOI
    BibTeX 
    @article{guo2024towards,
  title = {Towards transferable adversarial attacks on vision transformers for image classification},
  author = {Guo, Xu and Chen, Peng and Lu, Zhihui and Chai, Hongfeng and Du, Xin and Wu, Xudong},
  journal = {Journal of Systems Architecture},
  volume = {152},
  pages = {103155},
  year = {2024},
  doi = {10.1016/j.sysarc.2024.103155},
}