Search papers, labs, and topics across Lattice.
3
0
6
0
Diffusion models can navigate more reliably without retraining, thanks to a clever guidance method that keeps them on the training manifold.
Reference images can guide exposure correction with surprising accuracy, even without ground truth data or scene decomposition.
Multi-modal time series models can finally capture both magnitude and shape, thanks to a novel symbolic representation that discretizes continuous data into tokens.
