Denoising Network
The backbone of Diffusion Models is a denoising network, which is trained to gradually denoise data. While earlier works used a U-Net architecture, newer research has shown that Transformers can be used to achieve comparable or superior results. Modular Diffusion ships with both types of denoising network. Both are implemented in Pytorch and thinly wrapped in a Net module.
Future warning
The current denoising network implementations are not necessarily the most efficient or the most effective and are bound to change in a future release. They do, however, provide a great starting point for experimentation.
U-Net
U-Net implementation adapted from the The Annotated Diffusion Model. It takes an input with shape [b, c, h, w] and returns an output with shape [p, b, c, h, w].
Parameters
channels-> Sequence of integers representing the number of channels in each layer of the U-Net.labels(default0) -> Number of unique labels in .parameters(default1) -> Number of output parametersp.hidden(default256) -> Hidden dimension.heads(default8) -> Number of attention heads.groups(default16) -> Number of groups in the group normalization layers.
Example
from diffusion.net import UNet
net = UNet(channels=(3, 64, 128, 256), labels=10)Transformer
Transformer implementation adapted from the Peebles & Xie (2022)
(adaptive layer norm block). It takes an input with shape [b, l, e] and returns an output with shape [p, b, l, e].
Parameters
input-> Input embedding dimensione.labels(default0) -> Number of unique labels in .parameters(default1) -> Number of output parametersp.depth(default256) -> Number of transformer blocks.width(default256) -> Hidden dimension.heads(default8) -> Number of attention heads.
Example
from diffusion.net import Transformer
net = Transformer(input=x.shape[2])If you spot any typo or technical imprecision, please submit an issue or pull request to the library's GitHub repository .