Create a 1D, 2D, or 3D convolution module.

Zero out the parameters of a module and return it.

Evaluate a function without caching intermediate activations, allowing for reduced memory at the expense of extra compute in the backward pas

Make a standard normalization layer. :param channels: number of input channels. :return: an nn.Module for normalization.

Create a linear module.

Create sinusoidal timestep embeddings. :param timesteps: a 1-D Tensor of N indices, one per batch element. These may be

order: x -> Self Attn -> Cross Attn -> attn_s x -> Temp Self Attn -> attn_t x' = x + attn_s + attn_t

Inputs: - img_dir: str - The folder containing input images - output_dir: str - The workspace directory

Create a 1D, 2D, or 3D average pooling module.

Generate a bivariate isotropic or anisotropic Gaussian kernel. In the isotropic mode, only `sig_x` is used. `sig_y` and `theta` is ignored. Ar

decode frame by frame

encode frame by frame

This matches the implementation in Denoising Diffusion Probabilistic Models: From Fairseq. Build sinusoidal embeddings. This matches