Chen-Hao Chao
Ph.D. in CS @ University of Toronto
Hello, I'm a Computer Science Ph.D. student at the University of Toronto (UofT), advised by Prof. Rahul G. Krishnan. I am honored to be a recipient of the Mary H. Beatty Fellowship. I completed my master's and bachelor's degrees in Computer Science at National Tsing Hua University (NTHU). I collaborated on research projects with Prof. Chun-Yi Lee, visited Prof. Zsolt Kira's lab at Georgia Tech, and interned at NVIDIA and MediaTek.
My current research explores efficient pre-training methods and scaling behavior of diffusion language models, developing approaches that improve training and sampling efficiency. More broadly, my works cover probabilistic modeling and generative AI, spanning both discrete and continuous generative methods (e.g., score-based and flow-based models) with applications to reinforcement learning, visual domain adaptation, and biological data visualization. I built:
My current research explores efficient pre-training methods and scaling behavior of diffusion language models, developing approaches that improve training and sampling efficiency. More broadly, my works cover probabilistic modeling and generative AI, spanning both discrete and continuous generative methods (e.g., score-based and flow-based models) with applications to reinforcement learning, visual domain adaptation, and biological data visualization. I built:
- MDM-Prime (v1, v2): A scalable diffusion language model that allows partial word editings.
- EBFlow/MEow: A flow-based model that enables energy-based reinterpretation and RL applications.
- QCSBM: An architecturally flexible diffusion model that satisfies the conservative property.
- DLSM: A classifier-guidance diffusion model enhanced by likelihood score matching.
latest posts [full list]
selected publications [full list]
- TPAMIRainbow UDA: Combining Domain Adaptive Models for Semantic Segmentation TasksIEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2023