Changyue Liao

I received my M.S. from Zhejiang University. I work on LLM systems, with particular expertise in RL systems.

Publications

Ratel: Optimizing Holistic Data Movement to Fine-tune 100B Model on a Consumer GPU.

Changyue Liao, Mo Sun, Zihan Yang, Jun Xie, Kaiqi Chen, Binhang Yuan, Fei Wu, Zeke Wang.

IEEE International Conference on Data Engineering (ICDE), 2025 (PDF)

SmartReduce: SmartSwitch-Aware SmartNIC-Centric Sparse AllReduce

Yingtao Li, Hongjing Huang, Changyue Liao, Zihan Yang, Mo Sun, Xueying Zhu, Jie Zhang, Bo Zhao, Zeke Wang.

International Conference for High Performance Computing, Networking, Storage, and Analysis (SC), 2026

DisDP: Disaggregating Compute, Network, and Storage for Model-Sharded Data-Parallel Training.

Mo Sun, Zihan Yang, Changyue Liao, Yingtao Li, Jie Zhang, Kaiqi Chen, Fei Wu, Zeke Wang.

International Symposium on Computer Architecture (ISCA), 2026 (PDF)

Selected projects that acknowledge my support and contributions.

VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo — acknowledged for contributions to the VeOmni project.
SAIL-VL2 Technical Report — acknowledged for valuable support and contributions.

Ratel (LoHan)

A low-cost, high-performance system for fine-tuning large models; the LoHan project now hosts the Ratel artifact for ICDE 2025.

ASC Student Supercomputer Challenge

Projects developed for the ASC competition.

Visitor locations are estimated from IP addresses by MapMyVisitors and may be approximate.