Pretrained models of UVCGAN on the SLATS dataset

The files are pretrained network weights of UVCGAN on the Simple Liquid Argon Track Samples (SLATS) data set. The SLATS data set contains two domains, labeled fake and real, each populated by a variant of a LArTPC detector simulation used in the ProtoDUNE-SP experiment. UVCGAN is a deep neural network proposed for unpaired image translation. The files here are pretrained UVCGAN generators (or translators) that translate between the two domains of SLATS: net_gen_ab.pth translates fake to real and net_gen_ba.pth translate real to fake. The config.json contains the parameters needed to evaluate pretrained UVCGAN generators on SLATS.