← back to paper
arxiv: 2606.17412 · 2 revisions
Enhancing Pathological VLMs with Cross-scale Reasoning