The field of computer vision has seen significant advancements with the advent of convolutional neural networks (CNNs) and more recently vision transformers (ViTs), which have shown remarkable performance across diverse computer vision tasks in terms of prediction accuracy. In recent years, the focus has shifted towards improving the calibration of neural networks, a concept referring to the accuracy with which the model’s predictive scores reflect its predictive uncertainty. This research is focused on finding methods for improving calibration of modern neural networks.