Shaobo Wang (王少博)

Ditan Park, Beijing, 2026
Under old cypresses,
beside red-tiled eaves.

I am currently a Ph.D. student at Shanghai Jiao Tong University, under the supervision of Prof. Linfeng Zhang. Previously, I was a master’s student of ReThinkLab at SJTU, supervised by Prof. Junchi Yan.

I study data. My work centers on data for foundation models, especially at the pre-training and mid-training stages. Currently, I focus on coding and general agent capabilities at Moonshot Kimi. Previously, I worked on pre-training and mid-training at the Qwen Team. My research interests include:

Data selection and mixture: Maximizing data utility, including both quality and diversity (Qwen3/3.5, OPUS, Data Whisperer).
Data distillation and synthesis: Creating unseen data to break the data wall (Qwen3.6/3.7, NCFM, InfoUtil, Socratic-Zero).
Data evaluation: Building faithful evaluation for modeling real-world capabilities (EssenceBench, OPUS).

I have been fortunate to work with outstanding researchers, particularly Dr. Dayiheng Liu and Xingzhang Ren at the Qwen Team. During my internship at Qwen, I also collaborated with Huiqiang Jiang, Fei Huang, and Junyang Lin. Academically, I also collaborate closely with Dr. Conghui He at Shanghai AI Laboratory and Prof. Xuming Hu at HKUST(GZ). I am currently on the job market and seeking research scientist positions.

News

2026	My paper OPUS was accepted as an oral presentation at ICML 2026.
2025	I was selected as a Tencent Hunyuan Scholar in 2025, one of 23 recipients.
2025	My paper NCFM received full scores at CVPR 2025 and was selected as a Highlight paper.

Selected Publications

* denotes the equal contribution.

ICML oral

OPUS: Towards Efficient and Principled Data Selection in Large Language Model Pre-training in Every Iteration

Shaobo Wang , Xuan Ouyang , Tianyi Xu , Yuzheng Hu , Jialin Liu , Guo Chen , Tianyu Zhang , Junhao Zheng , and 4 more authors

In International Conference on Machine Learning , 2026

arXiv Bib HTML PDF Link

@inproceedings{wang2026opus,
  title = {OPUS: Towards Efficient and Principled Data Selection in Large Language Model Pre-training in Every Iteration},
  author = {Wang, Shaobo and Ouyang, Xuan and Xu, Tianyi and Hu, Yuzheng and Liu, Jialin and Chen, Guo and Zhang, Tianyu and Zheng, Junhao and Yang, Kexin and Ren, Xingzhang and Liu, Dayiheng and Zhang, Linfeng},
  booktitle = {International Conference on Machine Learning},
  year = {2026},
  eprint = {2602.05400},
  archiveprefix = {arXiv},
  url = {https://openreview.net/forum?id=FEfuE1mAB6},
}

ICLR

Grounding and Enhancing Informativeness and Utility in Dataset Distillation

Shaobo Wang , Yantai Yang , Guo Chen , Peiru Li , Kaixin Li , Yufa Zhou , Zhaorun Chen , and Linfeng Zhang

In The Fourteenth International Conference on Learning Representations , 2026

arXiv Bib PDF Link

@inproceedings{wang2026grounding,
  title = {Grounding and Enhancing Informativeness and Utility in Dataset Distillation},
  author = {Wang, Shaobo and Yang, Yantai and Chen, Guo and Li, Peiru and Li, Kaixin and Zhou, Yufa and Chen, Zhaorun and Zhang, Linfeng},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year = {2026},
  eprint = {2601.21296},
  archiveprefix = {arXiv},
  url = {https://openreview.net/forum?id=ThsYRbpv2F}
}

ICLR

Rethinking LLM Evaluation: Can We Evaluate LLMs with 200x Less Data?

Shaobo Wang* , Cong Wang* , Wenjie Fu* , Yue Min , Mingquan Feng , Isabel Guan , Xuming Hu , Conghui He , and 6 more authors

In The Fourteenth International Conference on Learning Representations , 2026

arXiv Bib PDF Code Link

@inproceedings{wang2026rethinking,
  title = {Rethinking LLM Evaluation: Can We Evaluate LLMs with 200x Less Data?},
  author = {Wang*, Shaobo and Wang*, Cong and Fu*, Wenjie and Min, Yue and Feng, Mingquan and Guan, Isabel and Hu, Xuming and He, Conghui and Wang, Cunxiang and Yang, Kexin and Ren, Xingzhang and Huang, Fei and Liu, Dayiheng and Zhang, Linfeng},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year = {2026},
  eprint = {2510.10457},
  archiveprefix = {arXiv},
  url = {https://openreview.net/forum?id=lZlZjSxdio}
}

ACL main

Data Whisperer: Efficient Data Selection for Task-Specific LLM Fine-Tuning via Few-Shot In-Context Learning

Shaobo Wang , Xiangqi Jin , Ziming Wang , Jize Wang , Jiajun Zhang , Kaixin Li , Zichen Wen , Zhong Li , and 3 more authors

In Annual Meeting of the Association for Computational Linguistics , 2025

arXiv Bib HTML PDF Code Website

@inproceedings{wang2025datawhisperer,
  title = {Data Whisperer: Efficient Data Selection for Task-Specific LLM Fine-Tuning via Few-Shot In-Context Learning},
  author = {Wang, Shaobo and Jin, Xiangqi and Wang, Ziming and Wang, Jize and Zhang, Jiajun and Li, Kaixin and Wen, Zichen and Li, Zhong and He, Conghui and Hu, Xuming and Zhang, Linfeng},
  booktitle = {Annual Meeting of the Association for Computational Linguistics},
  year = {2025},
  eprint = {2505.12212},
  archiveprefix = {arXiv},
}

CVPR highlight

Dataset Distillation with Neural Characteristic Function: A Minmax Perspective

Shaobo Wang , Yicun Yang , Zhiyuan Liu , Chenghao Sun , Xuming Hu , Conghui He , and Linfeng Zhang

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2025

arXiv Bib HTML PDF Code News (Chinese)

@inproceedings{wang2025dataset,
  title = {Dataset Distillation with Neural Characteristic Function: A Minmax Perspective},
  author = {Wang, Shaobo and Yang, Yicun and Liu, Zhiyuan and Sun, Chenghao and Hu, Xuming and He, Conghui and Zhang, Linfeng},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year = {2025},
  eprint = {2502.20653},
  archiveprefix = {arXiv},
  news_zh = {https://mp.weixin.qq.com/s/VtIqPF_a098qAEvrTKbi6A},
}

ICLR

Gnothi Seauton: Empowering Faithful Self-Interpretability in Black-Box Transformers

Shaobo Wang , Hongxuan Tang , Mingyang Wang , Hongrui Zhang , Xuyang Liu , Weiya Li , Xuming Hu , and Linfeng Zhang

In International Conference on Learning Representations , 2025

arXiv Bib PDF Code Link

@inproceedings{wang2025gnothi,
  title = {Gnothi Seauton: Empowering Faithful Self-Interpretability in Black-Box Transformers},
  author = {Wang, Shaobo and Tang, Hongxuan and Wang, Mingyang and Zhang, Hongrui and Liu, Xuyang and Li, Weiya and Hu, Xuming and Zhang, Linfeng},
  booktitle = {International Conference on Learning Representations},
  year = {2025},
  eprint = {2410.21815},
  archiveprefix = {arXiv},
  url = {https://openreview.net/forum?id=UvMSKonce8}
}