Scholar
Qinsheng Zhang
Google Scholar ID: hKwlT6wAAAAJ
Research Scientist, Nvidia
Citations & Impact
Citations
2,912
H-index
16
i10-index
20
Publications
20
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Co-authored papers on using generative diffusion models for unnormalized probability density sampling and introduced DEIS, which uses exponential integrators to reduce sampling steps to 10-15. Led the development of NVIDIA's open-source COSMOS world model platform.
Research Experience
- Leads large-scale diffusion model training at NVIDIA, including implementing efficient file I/O, designing parallel strategies, and working on COSMOS world models and Picasso projects.
Education
- Bachelor’s degree from Shanghai Jiaotong University in 2019; Ph.D. in Robotics from Georgia Institute of Technology IRIM in 2023, advised by Yongxin Chen.
Background
- From Quzhou, currently a senior research scientist at NVIDIA's Deep Imagination Research Group, focusing on probabilistic modeling of high-dimensional data. His research approach centers on the representation, learning, and sampling of complex probability distributions.
Miscellany
- Interests include teaching robots to play soccer, even winning a Robo WORLDCUP.
Co-authors