Our great sponsors
-
WorkOS
The modern identity platform for B2B SaaS. The APIs are flexible and easy-to-use, supporting authentication, user identity, and complex enterprise features like SSO and SCIM provisioning.
-
perceiver-pytorch
Implementation of Perceiver, General Perception with Iterative Attention, in Pytorch
-
hyperstyle
Official Implementation for "HyperStyle: StyleGAN Inversion with HyperNetworks for Real Image Editing" (CVPR 2022) https://arxiv.org/abs/2111.15666
There were two short snippets of code that were allegedly stolen, as far as I know. They were shown in a reply to https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/1936. I know the latter piece was nearly identical weeks ago, and the former is apparently how every project using hypernetworks initializes them.
From a quick search, a big part of the other code also seems to be basic boilerplate? For example half the lines match exactly to https://github.com/lucidrains/perceiver-pytorch/blob/main/perceiver_pytorch/perceiver_pytorch.py
Hypernetworks have been around, and fine tuning the VAE for Stable Diffusion is no secret either. People have just simply found the means to how NovelAI's team were producing their results. Now that the knowledge is out there, it cannot be prevented as history has proven time and time again. What NovelAI *needs* to do is tighten their security, get the team together on a new business model / plan, and come up with an even better implementation to what they already have. Leaks happen, you have to pivot, it is what it is.