This repository contains the code for the paper Regularization of Inverse Problems: Deep Equilibrium Models versus Bilevel Learning. Here we compare Bilevel Learning and Deep Equilibrium (DEQ) methods.
The code is based on the paper Deep Equilibrium Models, and in particular to the code which is publicly available here.
The current implementation allows training on either MNIST or CelebA datasets. To start, select the folder with the dataset you want to run (more instructions are included inside each folder).