About Me 🌲
Hi, I am a third-year Ph.D. student at the University of New South Wales (UNSW Sydney), supervised by Dr. Dong Gong. Before that, I received my MPhil degree from the Australian Institute for Machine Learning (AIML), University of Adelaide, supervised by A/Prof. Qi Wu and Dr. Yuankai Qi. I obtained my Bachelor’s degree from Beijing Jiaotong University, where I was advised by Prof. Runmin Cong.
My research interests lie in multimodal learning and continual learning, with a focus on continually improving LLMs/LVLMs to acquire new knowledge while mitigating forgetting. I am particularly interested in dynamic mixture of experts (Dynamic MoE) architectures and parameter-efficient fine-tuning (PEFT) for building scalable and adaptive pre-trained models.
Currently, I focus on world modeling — building consistent, efficient, and controllable world models that an agent can perceive, imagine, and act in:
- Continual learning — continually improving AI through post-training and test-time training, toward self-evolving models that learn from their own experience.
- Memory and world state — sparse, dynamic architectures that represent and manage rich world state, toward long-horizon consistency with sub-linear memory growth.
- Unified multimodal models — a single model that understands, reasons about, and generates the physical world, where understanding and generation reinforce or conflict across modalities and tokens, toward controllable world simulation.
News 🔥
May 2026Recognized as an ICML 2026 Gold Reviewer!May 2026MoRAM is accepted to ICML 2026. Congratulations to Jeff! Thanks to all collaborators!
Continual learning via incremental sparse mixture of rank-1 associative memory experts.Feb 2026DyMoE is accepted to CVPR 2026. Thanks to all collaborators!
“On Token’s Dilemma” — Continual learning dynamic MoE for LVLMs with token-level filtering.Jun 2025SAME is accepted to ICCV 2025. Congratulations to Gengze! Thanks to all collaborators!
State-adaptive MoE for generic language-guided visual navigation.Dec 2024New preprint! Check CoDyRA on arXiv!
Dynamic rank-selective LoRA for knowledge retention in continual learning VLMs/LLMs.Dec 2024New preprint! Check MambaCL on arXiv!
Meta-learning selective state space models (Mamba) for efficient continual learning.Oct 2024Recognized as an ACM MM 2024 Outstanding Reviewer!Dec 2023WebVLN is accepted to AAAI 2024. Thanks to all collaborators!
Vision-and-language navigation on shopping websites.Jul 2023MG-VLN is accepted to ACM MM 2023. Thanks to all collaborators!
Backtracking to passed correct locations via first-person video grounding for navigation.
Research 🌴
CVPRCVPR 2026
PreprintPreprint
ICMLICML 2026
PreprintPreprint
ACM MMACM MM 2023
AAAIAAAI 2024
ICCVICCV 2025
