Implementation of <a href="https://arxiv.org/abs/2212.11972">Recurrent Interface Network (RIN)</a>, for highly efficient generation of images and video without cascading networks, in Pytorch. The author unawaredly reinvented the <a href="https://github.com/lucidrains/isab-pytorch">induced set-attention block</a> from the <a href="https://arxiv.org/abs/1810.00825">set transformers</a> paper. They also combine this with the self-conditioning technique from the <a href="https://arxiv.org/abs/2208.04202">Bit Diffusion paper</a>, specifically for the latents. The last ingredient seems to be a new noise function based around the sigmoid, which the author claims is better than cosine scheduler for larger images.
0 commit comments