Hao-Xuan Ma   马浩轩

Master Student

LAMDA, Nanjing University

Email: mahx@lamda.nju.edu.cn;
             hunterwrynn@gmail.com;
Google Scholar: Google Scholar Link
Github: https://github.com/Hunter-Wrynn
LinkedIn: https://www.linkedin.com/in/haoxuan-ma-a50bba28a

Biography

I am currently a first-year graduate student at LAMDA@Nanjing University, advised by Prof. Han-Jia Ye.

I currently focus on Foundation Models and Model Reuse.

Selected Publications

* denotes equal contribution. ♦ denotes project leader.
MMR-Bench: A Comprehensive Benchmark for Multimodal LLM Routing
Haoxuan Ma*, Guannan Lai*, Han-Jia Ye
European Conference on Computer Vision (ECCV), 2026
[Paper] [Code]
Dual-Latent Memory Routing for Vision-Language Reasoning
Hao-Xuan Ma, Jin-Fei Qi, YiCheng Xiao, Han-Jia Ye
International Conference on Machine Learning (ICML), 2026 Spotlight
[Paper] [Code]
Routed Distillation for Multimodal RLVR
Hao-Xuan Ma, Han-Jia Ye
arXiv preprint, 2026
[Paper] [Code]
Token-Disentangled Latent Test-Time Scaling for Vision-Language Reasoning
Hao-Xuan Ma, Yihao Liu, Yutao Sun, Yanting Miao, Mengyu Zhou, Han-Jia Ye
arXiv preprint, 2026
[Paper] [Code]
Inverse Reinforcement Learning with Dynamic Reward Scaling for LLM Alignment
Ruoxi Cheng*, Haoxuan Ma*, Weixin Wang*, Ranjie Duan, Xiaojun Jia
International Conference on Learning Representations (ICLR), 2026
[Paper] [Code]
EcoAlign: An Economically Rational Framework for Efficient LVLM Alignment
Ruoxi Cheng*, Haoxuan Ma*, Teng Ma, Hongyi Zhang
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
[Paper] [Code]
Lever LM: Configuring In-Context Sequence to Lever Large Vision Language Models
Xu Yang, Yingzhe Peng, Haoxuan Ma, Shuo Xu, Chi Zhang, Yucheng Han, Hanwang Zhang
Neural Information Processing Systems (NeurIPS), 2024
[Paper] [Code]

Experience

Awards

Misc