TaskSelect Repository for experiments with ViT-based architectures for modelling Bottom-Up-Top-Down vision processes.