Hanmei Yang (杨寒梅)
Research Scientist
hanmei@meta.com
About Me
Currently, I am a Research Scientist at Meta, working on low-precision LLM training. I received my Ph.D. in Electrical and Computer Engineering from the University of Massachusetts Amherst, advised by Professor Tongping Liu and Professor Hui Guan. Before that, I received my B.Eng. from Southern University of Science and Technology, advised by Professor Jin Zhang.
I'm broadly interested in building efficient systems for large model training. My work spans from memory management — making large models fit and train efficiently on hardware — to quantization, where I explore how to train models faster without sacrificing quality through low-precision formats (e.g., MXFP4 and NVFP4).
Publications
- ProTrain: Efficient LLM Training via Automatic Memory Management, Hanmei Yang, Jin Zhou, Yao Fu, Xiaoquan Wang, Ramine Roane, Hui Guan, Tongping Liu Annual Conference on Machine Learning and Systems (MLSys) 2026.
- SNIP: An Adaptive Mixed Precision Framework for Subbyte Large Language Model Training, Yunjie Pan, Yongyi Yang, Hanmei Yang, Scott Mahlke ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) 2026.
- An Empirical Study of Microscaling Formats for Low-Precision LLM Training, Hanmei Yang, Summer Deng, Amit Nagpal, Maxim Naumov, Mohammad Janani, Tongping Liu, Hui Guan IEEE Symposium on Computer Arithmetic (ARITH) 2025.
- Understanding and Alleviating Memory Issue in RLHF for LLMs, Jin Zhou, Hanmei Yang, Steven (Jiaxun) Tang, Mingcan Xiang, Hui Guan, Tongping Liu Machine Learning for Systems Workshop at NeurIPS 2024.
- MemPerf: Profiling Allocator-Induced Performance Slowdowns, Jin Zhou, Sam Silvestro, Steven (Jiaxun) Tang, Hanmei Yang, Hongyu Liu, Guangming Zeng, Bo Wu, Cong Liu, Tongping Liu Object-Oriented Programming, Systems, Languages & Applications (OOPSLA) 2023.
- NUMAlloc: A Faster NUMA Memory Allocator, Hanmei Yang, Xin Zhao, Jin Zhou, Wei Wang, Sandip Kundu, Bo Wu, Hui Guan, Tongping Liu International Symposium on Memory Management (ISMM) 2023.
- Deadlock prediction via generalized dependency, Jinpeng Zhou, Hanmei Yang, John Lange, Tongping Liu International Symposium on Software Testing and Analysis (ISSTA) 2022.
- CachePerf: A Unified Cache Miss Classifier via Hybrid Hardware Sampling, Jin Zhou, Jiaxun Tang, Hanmei Yang, Tongping Liu ACM SIGMETRICS 2022.
- Capturing Human Pose Using mmWave Radar, Guangzheng Li, Ze Zhang, Hanmei Yang, Jin Pan, Dayin Chen, Jin Zhang Pervasive Computing and Communications Workshops (PerCom Workshops) 2020.
Service
- External Reviewer: IPDPS 2023
- PC Member: IPDPS 2026, PPoPP 2023 AE, EuroSys 2024 AE