Postdoc, University of California, San Diego
3 papers at NeurIPS 2025
We enable vision-language models to efficiently encode and utilize multimodal multilingual knowledge through a lightweight, continuous memory mechanism.
