Normlize image by given mean and std.

Prepare sample for usage as network input.

Resize sample to given size (width, height).

Normlize image by given mean and std.

r""" ConvNeXt A PyTorch impl of : `A ConvNet for the 2020s` - https://arxiv.org/pdf/2201.03545.pdf Args: in_chans (int

Feature fusion block.

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

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

Feature fusion block.

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

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

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

A downsampling layer with an optional convolution. Parameters: channels: channels in the inputs and outputs. use_conv: a boo

Interpolation module.

r""" LayerNorm that supports two data formats: channels_last (default) or channels_first. The ordering of the dimensions in the inputs. channels_

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

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

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

Quoting from https://arxiv.org/abs/2302.05543: "Stable Diffusion uses a pre-processing method similar to VQ-GAN [11] to convert the entire da

Modification of https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/levit.py: ConvNorm such that ConvTranspose2

Gradient loss

Network for monocular depth estimation.

Network for monocular depth estimation.

Residual convolution module.

Residual convolution module.

A dictionary of running averages.

SILog loss (pixel-wise)

Uses a pretrained BERT tokenizer by huggingface. Vocab size: 30522 (?)

Uses a pretrained BERT tokenizer by huggingface. Vocab size: 30522 (?)

r""" ConvNeXt Block. There are two equivalent implementations: (1) DwConv -> LayerNorm (channels_first) -> 1x1 Conv -> GELU -> 1x1 Conv; all in (N

Quoting from https://arxiv.org/abs/2302.05543: "Stable Diffusion uses a pre-processing method similar to VQ-GAN [11] to convert the entire da

Cross entropy loss

Uses the CLIP transformer encoder for text (from Hugging Face)

Uses the CLIP transformer encoder for text (from Hugging Face)

to match AttnBlock usage