This repository is the official implementation of our ECCV 2024 paper "Tight and Efficient Upper Bound on Spectral Norm of Convolutional Layers" by Ekaterina Grishina, Mikhail Gorbunov and Maxim Rakhuba.

• bounds.py contains code for different bounds on the spectral norm of convolution. compute_tensor_norm_einsum and compute_tensor_norm are the implementations of our proposed method. For the other bounds, we used code from the official repositories of the corresponding papers.
• pretrained_net_norms.py contains example of the spectral norm computation for the pretrained ResNet18 and VGG.