Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
1
0
3
Counterintuitively, explicitly balancing normal and abnormal examples in training batches hurts the zero-shot diagnostic accuracy of medical vision-language models.