A class representing an image that can be lazily loaded from various sources. This class supports loading an image from a filepath, URL, a P

A Pydantic model representing the input control parameters for an operation, typically involving image processing. This model includes p

Convolution + IBNorm + ReLu

Normlize image by given mean and std.

Tracks peak CUDA memory usage for a block of code.

Represents a prompt with an associated weight. This class is used to define a text prompt with a corresponding numerical weight, indicat

A sequential module that passes timestep embeddings to the children that support it as an extra input.

Lambda is a wrapper around a callable object that allows it to be used as a PyTorch module.

A residual block that can optionally change the number of channels. :param channels: the number of input channels. :param emb_channels: t

A sequential module that passes timestep embeddings to the children that support it as an extra input.

A residual block that can optionally change the number of channels. :param channels: the number of input channels. :param emb_channels: t

The model can behave as an encoder (with only self-attention) as well as a decoder, in which case a layer of cross-attention is added between

Transformer block for image-like data. First, project the input (aka embedding) and reshape to b, t, d. Then apply standard transform

An attention block that allows spatial positions to attend to each other. Originally ported from here, but adapted to the N-d case. https

Feature fusion block.

Subclass torch's LayerNorm to handle fp16.

A single linear layer which does a 1x1 conv

Fully connection operations for target net Note: Weights & biases are different for different images in a batch, thus here we

CLIPTextEncoderL is the CLIP text encoder with the following parameters: embedding_dim=768 num_layers=12 num_attention_heads=12 f

A downsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determini

A downsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determini

Feature fusion block.

An upsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determinin

CoordConv layer as in the paper.

Interpolation module.

Residual block recommended in: http://torch.ch/blog/2016/02/04/resnets.html

Residual Dense Block. Used in RRDB block in ESRGAN. Args: num_feat (int): Channel number of intermediate features. num_grow_

Resize sample to given size (width, height).

A Module that sets a context value when executed. The context need to pre exist in the context provider. #TODO Is there a way to create the c

An upsampling layer with an optional convolution. :param channels: channels in the inputs and outputs. :param use_conv: a bool determinin

A named tuple describing a ResNet block.