I am currently with the HY LLM Team as the principal researcher, where I am responsible for the mid-training stage of large language models. Our work focuses on understanding how synthetic data can be effectively used to scale LLM capabilities.
Previsouly, I am the contributor of Qwen 2.5&3 Series. Graph-Reasoner is a group of works who related to graph problem reasoning, you are welcome to visit it!
Research
We are looking for self-motivated interns with strong hands-on skills who are interested in LLM training, data-centric AI, and large-scale experiments.
We are particularly interested in several open questions:
- How can we improve the diversity of synthetic data?
- What proportion of synthetic data is optimal?
- How can synthetic data scale effectively?
- Can synthetic data replace real data?
If you are interested in working on data scaling laws, synthetic data generation, and mid-training strategies for LLMs, feel free to reach out.
News
- [12/2025] 🎉Two papers accepted by ICLR 2026.
- [09/2025] 🎉One paper accepted by Neurips 2025.
- [07/2025] 🎉Holds a tutorial about role-playing in IJCAI 2025.
- [05/2025] 🎉One paper accepted by COLM 2025.
- [05/2025] 🎉One paper accepted by KDD 2025 and Three papers accepeted by ACL 2025.
- [02/2025] 🎉Awarded 2024 Baidu Scholarship🎉
- [11/2024] 🎉One paper accepted by COLING 2025 (Oral）.
- [09/2024] 🎉🎉Two papers accepted by EMNLP 2024.
- [07/2024] I intern in Qwen Foundation Model Team, focusing on pre-training for improving llm reasoning.
- [07/2024] Our thoughts of Role-Playing are in this survey: The Oscars of AI Theater: A Survey on Role-Playing with Language Models.
- [05/2024] Severed as a reviewer at EMNLP 2024.
- [05/2024] 🎉One paper accepted by KDD 2024.
- [02/2024] 🥳🥳 Two recent works are avaliable at Arxiv: COMEDY and GraphWiz.
- [12/2023] 🥳🥳 Two recent works are avaliable at Arxiv: MathOctopus and LLaMA-Probing.
- [12/2023] Severed as a reviewer at NAACL and ACL 2024.
- [10/2023] 🎉🎉🎉 Three papers accepted by EMNLP 2023.
- [09/2023] Severed as a reviewer at T-PAMI.
- [05/2023] 🎉🎉Two papers accepted by ACL 2023.
Selected Publications
Awards
Services
- Invited Talk at Shanghai University Of Engineering Science
- Reviewer for ACMMM, ACL, EMNLP, AAAI, NeurIPS, NAACL, ICLR, T-PAMI and COLM.
Educations
- 2022.09 - Now, Phd Candidate, DSA, HKUST.
- 2019.09 - 2022.06, Master, Science of Computer Enginering, Peking University.
- 2015.09 - 2019.06, Undergraduate, Beijing University of Posts and Telecommunications.
Miscellaneous
LOL!!! Please feel free to contact me if you have any questions or suggestions: [email protected].