Search papers, labs, and topics across Lattice.
3
0
6
Token-level Mixture-of-Experts, directly ported from LLMs, can actually *hurt* autonomous driving performance in VLA models; SAMoE-VLA fixes this with scene-adaptive expert selection, achieving SOTA results with fewer parameters.
GloPath, a new foundation model trained on over a million glomeruli, achieves state-of-the-art performance in glomerular lesion assessment and uncovers novel links between kidney morphology and patient outcomes.
Achieve seamless, high-quality super-resolution on arbitrary-sized images with a diffusion model, sidestepping the limitations of fixed-size training and patch-based artifacts.
