Publications

2025

1. ICCV25 Highlight

   

   Rep-MTL: Unleashing the Power of Representation-level Task Saliency for Multi-Task Learning

   Zedong Wang, Siyuan Li, and Dan Xu

   ICCV (Highlight, Top 2.3%), 2025

   HTML PDF Code

2. 

   MergeVQ: A Unified Framework for Visual Generation and Representation with Disentangled Token Merging and Quantization

   Siyuan Li, Luyuan Zhang, Zedong Wang, and 8 more authors

   CVPR, 2025

   HF Daily Top-1 PDF Code

   MergeVQ ranked top 1 at Hugging Face Daily Papers on Apr 3rd, 2025

3. 

   Taming LLMs by Scaling Learning Rates with Gradient Grouping

   Siyuan Li^*, Juanxi Tian^*, Zedong Wang^*, and 4 more authors

   ACL (Main), 2025

   HF Daily Top-5 PDF Code

   SGG ranked top 5 at Hugging Face Daily Papers on June 3rd, 2025

4. 

   Prior-guided Hierarchical Harmonization Network for Efficient Image Dehazing

   Xiongfei Su, Siyuan Li, Yuning Cui, and 7 more authors

   AAAI, 2025

   PDF

2024

1. 

   MogaNet: Multi-order Gated Aggregation Network

   Siyuan Li^*, Zedong Wang^*, Zicheng Liu, and 6 more authors

   ICLR, 2024

   HTML PDF Code

2. 

   Unveiling the Backbone-Optimizer Coupling Bias in Visual Representation Learning

   Siyuan Li^*, Juanxi Tian^*, Zedong Wang^*, and 6 more authors

   arXiv preprint arXiv:2410.06373, 2024

   Bib PDF Code

   @article{li2024bocb,
     title = {Unveiling the Backbone-Optimizer Coupling Bias in Visual Representation Learning},
     author = {Li, Siyuan and Tian, Juanxi and Wang, Zedong and Zhang, Luyuan and Liu, Zicheng and Jin, Weiyang and Liu, Yang and Sun, Baigui and Li, Stan Z.},
     journal = {arXiv preprint arXiv:2410.06373},
     year = {2024},
   }

3. 

   Short-Long Convolutions Help Hardware-Efficient Linear Attention to Focus on Long Sequences

   Zicheng Liu, Siyuan Li , Li Wang, and 3 more authors

   ICML, 2024

   PDF Code

4. 

   VQDNA: Unleashing the Power of Vector Quantization for Multi-Species Genomic Sequence Modeling

   Siyuan Li^*, Zedong Wang^*, Zicheng Liu, and 5 more authors

   ICML, 2024

   PDF Code

5. 

   SemiReward: A General Reward Model for Semi-supervised Learning

   Siyuan Li, Weiyang Jin, Zedong Wang, and 4 more authors

   ICLR, 2024

   PDF Code

6. 

   LongVQ: Long Sequence Modeling with Vector Quantization on Structured Memory

   Zicheng Liu , Li Wang, Siyuan Li, and 3 more authors

   IJCAI, 2024

   PDF Code

7. 

   Switch EMA: A Free Lunch for Better Flatness and Sharpness

   Siyuan Li, Zicheng Liu, Juanxi Tian, and 9 more authors

   arXiv preprint arXiv:2211.03295, 2024

   Bib PDF

   @article{li2024SEMA,
     title = {Switch EMA: A Free Lunch for Better Flatness and Sharpness},
     author = {Li, Siyuan and Liu, Zicheng and Tian, Juanxi and Wang, Ge and Wang, Zedong and Jin, Weiyang and Wu, Di and Tan, Cheng and Lin, Tao and Liu, Yang and Sun, Baigui and Li, Stan Z.},
     journal = {arXiv preprint arXiv:2211.03295},
     year = {2024},
   }

2023

1. 

   Masked Modeling for Self-supervised Representation Learning on Vision and Beyond

   Siyuan Li, Luyuan Zhang, Zedong Wang, and 8 more authors

   arXiv preprint arXiv:2401.00897, 2023

   Bib PDF Code

   @article{li2023masked,
     title = {Masked Modeling for Self-supervised Representation Learning on Vision and Beyond},
     author = {Li, Siyuan and Zhang, Luyuan and Wang, Zedong and Wu, Di and Wu, Lirong and Liu, Zicheng and Xia, Jun and Tan, Cheng and Liu, Yang and Sun, Baigui and Li, Stan Z.},
     journal = {arXiv preprint arXiv:2401.00897},
     year = {2023},
   }

2. 

   OpenSTL: A Comprehensive Benchmark of Spatio-Temporal Predictive Learning

   Cheng Tan, Siyuan Li, Zhangyang Gao, and 5 more authors

   NeurIPS, 2023

   PDF Code

2022

1. 

   OpenMixup: Open Mixup Toolbox and Benchmark for Visual Representation Learning

   Siyuan Li^*, Zedong Wang^*, Zicheng Liu, and 2 more authors

   arXiv preprint arXiv:2209.04851, 2022

   Bib PDF Code

   @article{li2022openmixup,
     title = {OpenMixup: Open Mixup Toolbox and Benchmark for Visual Representation Learning},
     author = {Li, Siyuan and Wang, Zedong and Liu, Zicheng and Wu, Di and Li, Stan Z.},
     journal = {arXiv preprint arXiv:2209.04851},
     year = {2022},
   }

2021

1. 

   Boosting Discriminative Visual Representation Learning with Scenario-Agnostic Mixup

   Siyuan Li, Zicheng Liu, Zedong Wang, and 3 more authors

   arXiv preprint arXiv:2111.15454, 2021

   Bib PDF Code

   @article{li2021samix,
     title = {Boosting Discriminative Visual Representation Learning with Scenario-Agnostic Mixup},
     author = {Li, Siyuan and Liu, Zicheng and Wang, Zedong and Wu, Di and Liu, Zihan and Li, Stan Z},
     journal = {arXiv preprint arXiv:2111.15454},
     year = {2021},
   }