Scholar
Hanze Dong
Google Scholar ID: g9WLzWoAAAAJ
Microsoft Research
Citations & Impact
Citations
2,552
H-index
22
i10-index
29
Publications
20
Co-authors
44
list available
Publications
Resume (English only)
Academic Achievements
- Multiple papers accepted at top conferences such as NeurIPS, ICML, ACL, and EMNLP; Best Demo Award at NAACL 2024; Successfully defended his PhD thesis on November 30th, 2023; Involved in the development of the LMFlow framework.
Research Experience
- Senior Researcher at Microsoft Research and a founding member of the Singapore research lab. Previously, he was a Research Scientist at Salesforce Research. He serves as the managing editor of JMLR and has published in leading machine learning journals and conferences.
Education
- PhD: Department of Mathematics, HKUST, supervised by Professor Tong Zhang; BSc: Fudan University, Mathematics
Background
- His research interests include the reproducibility and interpretability of modern foundation models, post-training and alignment, generative modeling, and Monte Carlo sampling. He is also a core author of several influential open-source post-training GitHub packages. His theoretical work explores the core principles of generalization and optimization in foundation models, encompassing theory-guaranteed training algorithms and the analysis of diffusion-like dynamics.
Miscellany
- Open to research collaborations, especially in post-training, RL, and diffusion models. Can be reached via email.
