Mengzhao Chen 陈锰钊

Second-year Ph.D. Student

Department of Computer Science, The University of Hong Kong

Email: chenmnz@connect.hku.hk

WeChat ID: chenmnz1 (添加请注明来意)
[Github] [Scholar]

About Me [back top]

I am currently a second-year Ph.D. student in the University of Hong Kong, fortunately supervised by Prof. Ping Luo at the MMLAB@HKU. Earlier, I received the master and bachelor degree from MAC@XMU, Xiamen University, and advised by Prof. Rongrong Ji.

My research interests are to develop efficient models, including both vision and language models. Recently, I focus on develop efficient training and inference algorithm for Large Language Models (LLMs).

I am actively looking for academic collaboration, fell free to contact me if you are interested.

Latest News [back top]

[2026.07]: PrefixQuant is accepted by TPAMI.
[2026.05]: QAT Scaling Law and INT vs. FP are accepted by ICML, and QAT Scaling Law is selected as Spotlight paper.
[2025.11]: Awarded the Bytedance Scholarship
(20W RMB, 20 students in China and Singapore)
[2025.05]: EfficientQAT is accepted by ACL Main.
[2024.10]: Awarded the 中国电子学会-腾讯博士生科研激励计划
(10W RMB, 17 PHD students in China)
[2024.09] Start the PHD journal at HKU-MMLab
[2024.01]: Two paper are accepted by ICLR 2024, including one Spotlight
[2023.10]: Awarded the National Scholarship (国家奖学金)
[2023.07]: Two paper are accepted by ICCV 2023
[2023.07]: One paper is accepted by IJCV
[2022.11]: One paper is accepted as oral presentation by AAAI 2023

Selected Publications [back top]

LLMs Pretrain

	Mengzhao Chen, Meng Wu, Hui Jin, Zhihang Yuan, Jing Liu, Chaoyi Zhang, Yunshui Li, Jie Huang, Jin Ma, Zeyue Xue, Zhiheng Liu, Xingyan Bin^✉, Ping Luo^✉ INT vs. FP: A Comprehensive Study of Fine-Grained Low-bit Quantization Formats International Conference on Machine Learning (ICML), 2026 [arXiv] [code ]
	Bohong Wu, Mengzhao Chen, Xiang Luo, Shen Yan, Qifan Yu, Fan Xia, Tianqi Zhang, Hongrui Zhan, Zheng Zhong, Xun Zhou, Siyuan Qiao, Xingyan Bin^✉ Parallel Loop Transformer for Efficient Test-Time Computation Scaling** [arXiv] (^* Equal Contribution)

LLMs QAT

	Mengzhao Chen, Chaoyi Zhang, Jing Liu, Yutao Zeng, Zeyue Xue, Zhiheng Liu, Yunshui Li, Jin Ma, Jie Huang, Xun Zhou^✉, Ping Luo^✉ Scaling Law for Quantization-Aware Training International Conference on Machine Learning (ICML, Spotlight), 2026 [arXiv]
	Mengzhao Chen, Wenqi Shao^✉, Peng Xu, Jiahao Wang, Peng Gao, Kaipeng Zhang, Yu Qiao, Ping Luo^✉ EfficientQAT: Efficient Quantization-Aware Training for Large Language Models Association for Computational Linguistics (ACL, Main), 2025 [arXiv] [code] [中文介绍]

LLMs PTQ

	Mengzhao Chen, Yi Liu, Jiahao Wang, Yi Bin, Wenqi Shao^✉, Ping Luo^✉ PrefixQuant: Eliminating Outliers by Prefixed Tokens for Large Language Models Quantization IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2026 [arXiv] [code ]
	Wenqi Shao^,Mengzhao Chen^, Zhaoyang Zhang, Peng Xu, Lirui Zhao, Zhiqian Li, Kaipeng Zhang, Peng Gao, Yu Qiao, Ping Luo^✉ OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models International Conference on Learning Representations (ICLR, Spotlight), 2024 [arXiv] [code ] [中文介绍] (^* Equal Contribution)

ViT Acceleration

	Mengzhao Chen,Wenqi Shao^✉, Peng Xu, Mingbao Lin, Kaipeng Zhang, Fei Chao, Rongrong Ji^✉, Yu Qiao, Ping Luo DiffRate : Differentiable Compression Rate for Efficient Vision Transformers International Conference on Computer Vision (ICCV), 2023 [arXiv] [code ] [中文介绍] [直播回放]
	Mengzhao Chen, Mingbao Lin, Ke Li, Yunhang Shen, Yongjian Wu, Fei Chao, Rongrong Ji^✉ CF-ViT: A General Coarse-to-Fine Method for Vision Transformer AAAI Conference on Artificial Intelligence (AAAI, Oral), 2023 [arXiv] [code ] [中文介绍]
	Mengzhao Chen, Mingbao Lin, Zhihang Lin, Yuxin Zhang, Fei Chao, Rongrong Ji^✉ SMMix: Self-Motivated Image Mixing for Vision Transformers International Conference on Computer Vision (ICCV), 2023 [arXiv] [code]
	Mingbao Lin^, Mengzhao Chen^, Yuxin Zhang, Ke Li, Yunhang Shen, Chunhua Shen, Rongrong Ji, Liujuan Cao^✉ Super Vision Transformer International Journal of Computer Vision (IJCV), 2023 [arXiv] [code] (^* Equal Contribution)