Search papers, labs, and topics across Lattice.
2
0
4
VISReg not only stabilizes embedding training but also achieves state-of-the-art performance with a fraction of the data used by competing methods.
Training on 500K automatically-curated ophthalmology instructions lets a vision-language model leapfrog general medical models in a specialized domain.
