Didi Zhu

@inproceedings{gao2026adavig,
  title     = {{Model Guides You How to Draw: Adaptive Visual Gating for Unified Multimodal Reasoning}},
  author    = {Gao, Wenxi and Lu, Guanxi and Zhu, Didi and Chen, Hao Mark and Deng, Quan and Wang, Zhican and Deng, Jiankang and Fan, Hongxiang},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year      = {2026},
  eprint    = {2607.10004},
  archivePrefix = {arXiv}
}

@article{zhu2026visualflip,
  title     = {{VisualFLIP: Do Predictions Depend on Task-Critical Visual Evidence in Multimodal Reasoning?}},
  author    = {Zhu, Didi and Chen, Changrui and Zafeiriou, Stefanos and Deng, Jiankang},
  journal   = {arXiv preprint arXiv:2606.07872},
  year      = {2026},
  eprint    = {2606.07872},
  archivePrefix = {arXiv}
}

@article{an2026llavaonevision2,
  title     = {{LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence}},
  author    = {An, Xiang and Xie, Yin and Tang, Feilong and Yan, Yunyao and Tan, Huajie and Zhu, Didi and Chen, Changrui and Zhao, Xiuwei and Qin, Bin and Yang, Kaicheng and Shen, Yifei and Zhang, Yuanhan and Zhang, Kaichen and Zhang, Wenkang and Cheng, Zheng and Zhang, Nansen and Wu, Chunsheng and Ge, Chunjiang and Ran, Zimin and Song, Dehua and Li, Chunyuan and Feng, Shikun and Hu, Ming and Chen, Zhangquan and Niu, Junbo and Li, Bo and Feng, Ziyong and Liu, Ziwei and Ge, Zongyuan and Deng, Jiankang},
  journal   = {arXiv preprint arXiv:2605.25979},
  year      = {2026},
  eprint    = {2605.25979},
  archivePrefix = {arXiv}
}

@inproceedings{tong2026alleviating,
  title     = {{Alleviating Sparse Rewards by Modeling Step-Wise and Long-Term Sampling Effects in Flow-Based GRPO}},
  author    = {Tong, Yunze and Liu, Mushui and Zhao, Canyu and Zhu, Didi and He, Wanggui and Zhang, Shiyi and Zhang, Hongwei and Zhang, Peng and Liu, Jinlong and Jiang, Hao},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2026},
  eprint    = {2602.06422},
  archivePrefix = {arXiv}
}

@inproceedings{zhao2026smora,
  title     = {{Each Rank Could be an Expert: Single-Ranked Mixture of Experts LoRA for Multi-Task Learning}},
  author    = {Zhao, Ziyu and Zhou, Yixiao and Zhang, Zhi and Zhu, Didi and Shen, Tao and Li, Zexi and Yang, Jinluan and Wang, Xuwu and Su, Jing and Kuang, Kun and Wei, Zhongyu and Wu, Fei and Cheng, Yu},
  booktitle = {Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)},
  year      = {2026},
  eprint    = {2501.15103},
  archivePrefix = {arXiv}
}

@article{tong2026infinityedit,
  title     = {{InfinityEdit: Infinite Video Editing with a Lightweight Edit-Ignition Adapter}},
  author    = {Tong, Yunze and Liu, Mushui and Zhao, Canyu and Zhang, Shiyi and Zhu, Didi and Zhang, Peng and He, Wanggui and Liu, Jinlong and Chen, Ying and Jiang, Hao and Huang, Pipei and Zheng, Bo},
  journal   = {arXiv preprint arXiv:2608.20910},
  year      = {2026},
  eprint    = {2608.20910},
  archivePrefix = {arXiv}
}

@misc{zhu2025llavaonevision15rl,
  title     = {{LLaVA-OneVision-1.5-RL: Unlocking Multimodal Reasoning via Lightweight Reinforcement Learning}},
  author    = {Zhu, Didi and Qu, Zhiyu and Chen, Zerui and Gkagkos, Polydefkis and An, Xiang and Li, Bo},
  howpublished = {\url{https://mvp-ai-lab.github.io/LLaVA-OneVision-1.5-RL/}},
  note      = {RL section of the LLaVA-OneVision-1.5 technical report, arXiv:2509.23661},
  year      = {2025}
}

@article{an2025llavaonevision15,
  title     = {{LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training}},
  author    = {An, Xiang and Xie, Yin and Yang, Kaicheng and Zhang, Wenkang and Zhao, Xiuwei and Cheng, Zheng and Wang, Yirui and Xu, Songcen and Chen, Changrui and Zhu, Didi and Wu, Chunsheng and Tan, Huajie and Li, Chunyuan and Yang, Jing and Yu, Jie and Wang, Xiyao and Qin, Bin and Wang, Yumeng and Yan, Zizhen and Feng, Ziyong and Liu, Ziwei and Li, Bo and Deng, Jiankang},
  journal   = {arXiv preprint arXiv:2509.23661},
  year      = {2025},
  eprint    = {2509.23661},
  archivePrefix = {arXiv}
}

@inproceedings{lu2025watch,
  title     = {{Watch Wider and Think Deeper: Collaborative Cross-modal Chain-of-Thought for Complex Visual Reasoning}},
  author    = {Lu, Wenting and Zhu, Didi and Shen, Tao and Zhu, Donglin and Ye, Ayong and Wu, Chao},
  booktitle = {NeurIPS 2025 Workshop on Multimodal Algorithmic Reasoning},
  year      = {2025},
  eprint    = {2601.02422},
  archivePrefix = {arXiv}
}

@inproceedings{tong2025decoding,
  title     = {{Decoding Correlation-Induced Misalignment in the Stable Diffusion Workflow for Text-to-Image Generation}},
  author    = {Tong, Yunze and Zhang, Fengda and Zhu, Didi and Xiao, Jun and Kuang, Kun},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  year      = {2025}
}

@article{tong2025noise,
  title     = {{Noise Projection: Closing the Prompt-Agnostic Gap Behind Text-to-Image Misalignment in Diffusion Models}},
  author    = {Tong, Yunze and Zhu, Didi and Hu, Zijing and Yang, Jinluan and Zhao, Ziyu},
  journal   = {arXiv preprint arXiv:2510.14526},
  year      = {2025},
  eprint    = {2510.14526},
  archivePrefix = {arXiv}
}

@inproceedings{yang2025mix,
  title     = {{Mix Data or Merge Models? Balancing the Helpfulness, Honesty, and Harmlessness of Large Language Model via Model Merging}},
  author    = {Yang, Jinluan and Jin, Dingnan and Tang, Anke and Shen, Li and Zhu, Didi and Chen, Zhengyu and Zhao, Ziyu and Wang, Daixin and Cui, Qing and Zhang, Zhiqiang and Zhou, Jun and Wu, Fei and Kuang, Kun},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year      = {2025},
  eprint    = {2502.06876},
  archivePrefix = {arXiv}
}

@inproceedings{dong2025erict,
  title     = {{ERICT: Enhancing Robustness by Identifying Concept Tokens in Zero-Shot Vision Language Models}},
  author    = {Dong, Xinpeng and Zhang, Min and Zhu, Didi and Jian, Ye Jun and Zhang, Keli and Zhou, Aimin and Wu, Fei and Kuang, Kun},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2025}
}

@inproceedings{feng2025zeroflow,
  title     = {{ZeroFlow: Overcoming Catastrophic Forgetting is Easier than You Think}},
  author    = {Feng, Tao and Li, Wei and Zhu, Didi and Yuan, Hangjie and Zheng, Wendi and Zhang, Dan and Tang, Jie},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2025},
  eprint    = {2501.01045},
  archivePrefix = {arXiv}
}

@inproceedings{huang2025learn,
  title     = {{Learn from Downstream and Be Yourself in Multimodal Large Language Model Fine-Tuning}},
  author    = {Huang, Wenke and Liang, Jian and Shi, Zekun and Zhu, Didi and Wan, Guancheng and Li, He and Du, Bo and Tao, Dacheng and Ye, Mang},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2025},
  eprint    = {2411.10928},
  archivePrefix = {arXiv}
}

@inproceedings{huang2025confident,
  title     = {{Be Confident: Uncovering Overfitting in MLLM Multi-Task Tuning}},
  author    = {Huang, Wenke and Liang, Jian and Wan, Guancheng and Zhu, Didi and Li, He and Shao, Jiawei and Ye, Mang and Du, Bo and Tao, Dacheng},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2025}
}

@inproceedings{zhu2025remedy,
  title     = {{REMEDY: Recipe Merging Dynamics in Large Vision-Language Models}},
  author    = {Zhu, Didi and Song, Yibing and Shen, Tao and Zhao, Ziyu and Yang, Jinluan and Zhang, Min and Wu, Chao},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2025}
}

@inproceedings{yang2025backdoor,
  title     = {{Mitigating the Backdoor Effect for Multi-Task Model Merging via Safety-Aware Subspace}},
  author    = {Yang, Jinluan and Tang, Anke and Zhu, Didi and Chen, Zhengyu and Shen, Li and Wu, Fei},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2025},
  eprint    = {2410.13910},
  archivePrefix = {arXiv}
}

@inproceedings{zhao2025merging,
  title     = {{Merging LoRAs like Playing LEGO: Pushing the Modularity of LoRA to Extremes Through Rank-Wise Clustering}},
  author    = {Zhao, Ziyu and Shen, Tao and Zhu, Didi and Li, Zexi and Su, Jing and Wang, Xuwu and Kuang, Kun and Wu, Fei},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2025},
  eprint    = {2409.16167},
  archivePrefix = {arXiv}
}

@inproceedings{li2025fedgucci,
  title     = {{FedGuCci: Making Local Models More Connected in Landscape for Federated Learning}},
  author    = {Li, Zexi and Lin, Jie and Li, Zhiqi and Zhu, Didi and Shen, Tao and Lin, Tao and Wu, Chao and Lane, Nicholas D.},
  booktitle = {Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)},
  year      = {2025},
  eprint    = {2402.18949},
  archivePrefix = {arXiv}
}

@article{shen2025scaling,
  title     = {{Will LLMs Scaling Hit the Wall? Breaking Barriers via Distributed Resources on Massive Edge Devices}},
  author    = {Shen, Tao and Zhu, Didi and Zhao, Ziyu and Li, Zexi and Wu, Chao and Wu, Fei},
  journal   = {arXiv preprint arXiv:2503.08223},
  year      = {2025},
  eprint    = {2503.08223},
  archivePrefix = {arXiv}
}

@inproceedings{zhu2024model,
  title     = {{Model Tailor: Mitigating Catastrophic Forgetting in Multi-modal Large Language Models}},
  author    = {Zhu, Didi and Sun, Zhongyi and Li, Zexi and Shen, Tao and Yan, Ke and Ding, Shouhong and Kuang, Kun and Wu, Chao},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2024},
  eprint    = {2402.12048},
  archivePrefix = {arXiv}
}

@inproceedings{zhu2024neural,
  title     = {{Neural Collapse Anchored Prompt Tuning for Generalizable Vision-Language Models}},
  author    = {Zhu, Didi and Li, Zexi and Zhang, Min and Yuan, Junkun and Liu, Jiashuo and Kuang, Kun and Wu, Chao},
  booktitle = {Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)},
  year      = {2024},
  eprint    = {2306.15955},
  archivePrefix = {arXiv}
}

@inproceedings{zhu2023universal,
  title     = {{Universal Domain Adaptation via Compressive Attention Matching}},
  author    = {Zhu, Didi and Li, Yinchuan and Yuan, Junkun and Li, Zexi and Kuang, Kun and Wu, Chao},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  year      = {2023}
}

@inproceedings{zhu2023generalized,
  title     = {{Generalized Universal Domain Adaptation with Generative Flow Networks}},
  author    = {Zhu, Didi and Li, Yinchuan and Shao, Yunfeng and Hao, Jianye and Wu, Fei and Kuang, Kun and Xiao, Jun and Wu, Chao},
  booktitle = {Proceedings of the ACM International Conference on Multimedia (MM)},
  year      = {2023},
  eprint    = {2305.04466},
  archivePrefix = {arXiv}
}

@inproceedings{tong2023quantitatively,
  title     = {{Quantitatively Measuring and Contrastively Exploring Heterogeneity for Domain Generalization}},
  author    = {Tong, Yunze and Yuan, Junkun and Zhang, Min and Zhu, Didi and Zhang, Keli and Wu, Fei and Kuang, Kun},
  booktitle = {Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)},
  year      = {2023},
  eprint    = {2305.15889},
  archivePrefix = {arXiv}
}

@article{li2022towards,
  title     = {{Towards Effective Clustered Federated Learning: A Peer-to-peer Framework with Adaptive Neighbor Matching}},
  author    = {Li, Zexi and Lu, Jiaxun and Luo, Shuang and Zhu, Didi and Shao, Yunfeng and Li, Yinchuan and Zhang, Zhimeng and Wang, Yongheng and Wu, Chao},
  journal   = {IEEE Transactions on Big Data},
  year      = {2022},
  eprint    = {2203.12285},
  archivePrefix = {arXiv}
}

@inproceedings{luo2021ensemble,
  title     = {{Ensemble Federated Adversarial Training with Non-IID Data}},
  author    = {Luo, Shuang and Zhu, Didi and Li, Zexi and Wu, Chao},
  booktitle = {IJCAI 2021 Workshop on Federated and Transfer Learning for Data Sparsity and Confidentiality (FTL-IJCAI)},
  year      = {2021},
  eprint    = {2110.14814},
  archivePrefix = {arXiv}
}