About me
I am an Ph.D. student in the Department of Computer Science and Engineering at University of Connecticut since 2021 spring. I obtained my B.S. degree in electrical engineering from Huazhong University of Science and Technology in 2018 and and M.S. degree in electrical engineering from University of Akransas in 2020.
I am generally interested in Large Language Model, efficient ML, secure and Trustworthy AI/ML, and ML system. My current research focuses are listed as follows:
Recent News
09/2024 Our paper LoT regularization for improving the generalization of model training has been accepted by NeurIPS 24. LoT is proved to be highly effective accross tasks such as LLM finetune, Game RL, Small LM pretraining, and Image Classification [Code].
05/2024 I’m starting a new internship at Adobe Research team, under the supervision of Yan Kang, Marian Petre, Dr. Yuchen Liu, Tobias Hinz, Zhe Lin. Research Topic: GenAI (text2image, text2video) efficiency
12/2023 I’m starting a new internship at Microsoft GenAI team, under the supervision of Abhishek Goswami, Dr. Shuohang Wang, Dr. Yelong Shen, Dr. Weizhu Chen. Research Topic: Representation Learning for Large Language Model
11/2023 Our paper MaxK-GNN on accelerating GNN training through MaxK nonlinearity & GPU kernel co-design has been accepted by ASPLOS 24. [Code].
10/2023 I accepted the Synchrony Fellowships from UConn CSE. Thanks UConn CSE![News]
09/2023 Our paper LinGCN on accelerating GCN private inference under Homomorphical Encryption Setting has been accepted by NEURIPS 23. [Code].
09/2023 We release Medusa, a easy-to-use framework which accelerates LLM generation through multiple light-weighted decoding head. No draft model needed! [Blog], [code].
07/2023 Our paper AQ2PNN on adaptive quantization in private inference under multi-party computation setting has been accepted by MICRO 23.
07/2023 Our paper Accel-GCN on graph learning acceleration on GPUs has been accepted by ICCAD 23. [Code].
07/2023 Our paper AutoReP on ReLU replacement for fast private inference under multi-party computation setting has been accepted by ICCV 23. [Code].
07/2023 I accepted the GE Fellowship from School of Engineering. Thanks School of Engineering!
05/2023 I accepted the Predoctoral Fellowship from UConn CSE. Thanks UConn CSE!
02/2023 Our paper PASNet on NAS for private inference acceleration under multi-party computation setting has been accepted by DAC 23. [Code].
01/2023 Our paper on GNN sparsification has been accepted by ICCD 23. [Code].
05/2022 I accepted the Taylor L. Booth predoctoral fellowship for top 1 scholarly achievements by UConn CSE. Thanks UConn CSE!
02/2022 Our paper on Transformer model acceleration has been accepted by DAC 2022.