@article{kang2026xema,title={Xema: Efficient Diffusion Serving through Fine-Grained Memory Management and Auto-Configuration},author={Kang, Xueze and Xiang, Guangyu and Li, Suyi and Wang, Yuxin and Shi, Shaohuai and Zhang, Lin and Chu, Xiaowen},journal={arXiv preprint arXiv:2607.11136},year={2026},}
KernelFlume: Elastic Core-Attention Scaling for Agentic Long-Context Decoding
Guangyu Xiang*, Xueze Kang*, Lin Zhang, Wenxiang Lin, Shaohuai Shi, Yuxin Wang, and Xiaowen Chu
@article{xiang2026kernelflume,title={KernelFlume: Elastic Core-Attention Scaling for Agentic Long-Context Decoding},author={Xiang, Guangyu and Kang, Xueze and Zhang, Lin and Lin, Wenxiang and Shi, Shaohuai and Wang, Yuxin and Chu, Xiaowen},journal={arXiv preprint arXiv:2606.29207},year={2026},}
Compass: Dissecting Communication and Computation Operators for Efficient LLM Training
Guangyu Xiang, Lin Zhang, Haoxuan Yu, Xinglin Pan, Shaohuai Shi, and Xiaowen Chu
In IEEE INFOCOM 2026 - IEEE Conference on Computer Communications, 2026
@inproceedings{xiang2026compass,title={Compass: Dissecting Communication and Computation Operators for Efficient LLM Training},author={Xiang, Guangyu and Zhang, Lin and Yu, Haoxuan and Pan, Xinglin and Shi, Shaohuai and Chu, Xiaowen},booktitle={IEEE INFOCOM 2026 - IEEE Conference on Computer Communications},pages={1--10},year={2026},doi={10.1109/INFOCOM59046.2026.11571341},}
2025
ElasWave: An Elastic-Native System for Scalable Hybrid-Parallel Training
Xueze Kang*, Guangyu Xiang*, Yuxin Wang, Hao Zhang, Yuchu Fang, Yuhang Zhou, Zhenheng Tang, Youhui Lv, Eliran Maman, Mark Wasserman, and 9 more authors
@article{kang2025elaswave,title={ElasWave: An Elastic-Native System for Scalable Hybrid-Parallel Training},author={Kang, Xueze and Xiang, Guangyu and Wang, Yuxin and Zhang, Hao and Fang, Yuchu and Zhou, Yuhang and Tang, Zhenheng and Lv, Youhui and Maman, Eliran and Wasserman, Mark and Zameret, Alon and Bian, Zhipeng and Chen, Shushu and Yu, Zhiyou and Wang, Jin and Wu, Xiaoyu and Zheng, Yang and Tian, Chen and Chu, Xiaowen},journal={arXiv preprint arXiv:2510.00606},year={2025},}