…Our model outperforms state-of-the-art models on both zero-shot and linear probing tasks for classifying new pathology images across 13 diverse patch-level datasets of 8 different sub-pathologies and cross-modal
4 PAPERS • NO BENCHMARKS YET