← back to paper
arxiv: 2604.21360 · 2 revisions
Prototype-Based Test-Time Adaptation of Vision-Language Models