lhh

Multimodal learning, vision-language models, and embodied AI. I’m interested in how post-training shapes representations and generalization across tasks and environments.