Research

Research Overview

My research focuses on the foundations of trustworthy and generalizable AI, with an emphasis on understanding how large models learn, generalize, and reason.

Current Interests

• Interpretability of Large Language Models

• LLMs and Linguistics

• Compositional Generalization

• Multi-agent Systems

Selected Publications

1. Jingwen Fu, Nanning Zheng. On the Statistical Mechanisms of Distributional Compositional Generalization. International Conference on Machine Learning (ICML), 2025.

2. Jingwen Fu, Tao Yang, Yuwang Wang, Yan Lu, Nanning Zheng. Breaking through the Learning Plateaus of In-context Learning in Transformer. International Conference on Machine Learning (ICML), 2024.

3. Jingwen Fu, Zhizheng Zhang, Dacheng Yin, Yan Lu, Nanning Zheng. Learning Trajectories are Generalization Indicators. Advances in Neural Information Processing Systems (NeurIPS), 2023.

4. Bowen Wang, Jingwen Fu, Haoran Zhang, Nanning Zheng, Wei Chen. Closing the Gap Between the Upper Bound and the Lower Bound of Adam's Iteration Complexity. Advances in Neural Information Processing Systems (NeurIPS), 2023.

A fuller list of papers and citations is on Google Scholar.