We explore the interplay between memorization and generalization in neural networks. Includes Memorization-Aware Training (MAT), a novel framework to mitigate the adverse effects of memorization and ...
Python 3.8 and PyTorch 1.13 and CUDA 11.8. Firstly, we compare the generated images by the theoretical optimum and state-of-the-art diffusion model (EDM). The experiments are run on a single A100 GPU.