CERTH-ITI-IDT/T-TAME-models

T-TAME: Trainable Attention Mechanism for Explaining Convolutional Networks and Vision Transformers: Supplemtary Material

This repository hosts the trained T-TAME attention mechanism for VGG-16, ViT-B-16, and ResNet-50 and the L-CAM method for the VGG-16 and ResNet-50 classifiers, used for comparisons. The checkpoints are bundled in the repository using git-lfs. It is a submodule of the main repository, which contains the code for the T-TAME method. The main repository can be found here.