LAMDA-CL Lab is part of the LAMDA group at Nanjing University. Our research focuses on continual learning, foundation model adaptation, and self-evolving AI, with the goal of building intelligent systems that can continuously acquire knowledge, adapt to changing environments, and improve themselves over time.

We are interested in how models can learn, adapt, reuse, and evolve across tasks, domains, and modalities. Our work spans both fundamental learning principles and practical methodologies for developing adaptive, efficient, and increasingly capable AI systems.

Members of LAMDA-CL actively contribute to leading AI venues including CVPR, ICML, NeurIPS, ICLR, TPAMI, and IJCV.

Our current research interests include:

• Continual and Lifelong Learning: Learning continuously from evolving data, tasks, and environments while retaining and reusing previously acquired knowledge.
• Foundation Model Adaptation and Reuse: Efficiently adapting, extending, and composing pre-trained and foundation models for new tasks and scenarios.
• Self-Evolving AI: Developing models that can autonomously refine their knowledge, representations, capabilities, and learning strategies through interaction with data, feedback, tools, and environments.
• Multimodal Learning and Adaptation: Studying continual and adaptive learning for vision-language models and multimodal foundation models.
• Efficient Learning: Enabling effective model adaptation under limited data, memory, supervision, and computational resources.
• Toolkits and Benchmarks: Building open-source platforms and benchmarks for reproducible research in continual, adaptive, and self-evolving learning.

We develop and maintain open-source toolkits for continual and adaptive learning:

• PyCIL: A Python toolbox for class-incremental learning.
• PILOT: A toolkit for continual learning with pre-trained models.
• C3Box: A CLIP-based class-incremental learning toolbox.
• Prism: A toolbox for multimodal continual instruction tuning.

We are always looking for motivated PhD students, postdocs, Master's students, and visiting interns who are interested in continual learning, foundation models, multimodal learning, and self-evolving AI.

If you are passionate about building intelligent systems that can continually learn, adapt, and evolve, we welcome you to join us.