I am a research master student (Jan. 2025 – now) at Mila - Quebec AI Institute and Concordia University, working on learned optimization for vision and language modeling with Prof. Eugene Belilovsky.

I am also actively exploring vision-language models, with a particular interest in their scaling philosophy, grounded real-world understanding and reasoning capabilities.

Feel free to reach out for interesting discussions and potential collaborations.

Studying AI is a continual pursuit of deeper understanding and self-transcendence.

Github / X / Google scholar / Email

Papers

Efficient Long-Horizon Learning for Learned Optimization. Xiaolong Huang, Benjamin Therien, James Harrison, Eugene Belilovsky. Preprint. paper / project page / code
MuLoCo: Muon is a practical inner optimizer for DiLoCo. Benjamin Therien, Xiaolong Huang, Aaron Defazio, Irina Rish, Eugene Belilovsky. ICML'26. paper / project page / code
One step learning, one step review. Xiaolong Huang, Qiankun Li, Xueran Li, Xuesong Gao. AAAI'24. paper / code
Data-efficient masked video modeling for self-supervised action recognition. Qiankun Li, Xiaolong Huang, Zhifan Wan, Lanqing Hu, Shuzhe Wu, Jie Zhang, Shiguang Shan, Zengfu Wang. ACM MM'23. paper
2nd place solution to Google Universal Image Embedding. Xiaolong Huang, Qiankun Li. ILR Workshop, ECCV'22. Oral. paper / code / competition

Academic Service

Reviewer
NeurIPS, AAAI