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

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

Ramdom clip x to win_length. Args: x (tensor) : (B, c_in, T, n_bins). cond (tensor) : (B, T, H). x_len (te

Create a linear module.

Zero out the parameters of a module and return it.

:param x: img of size (bs, c, h, w) :return: n img crops of size (n, bs, c, kernel_size[0], kernel_size[1])

:param x: img of size (bs, c, h, w) :return: n img crops of size (n, bs, c, kernel_size[0], kernel_size[1])

r"""Load list of class labels and return text embeddings

r"""Process audio and/or text per batch

r"""Process audio and/or text per batch

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

r"""Compute similarity between text and audio embeddings

r"""Load list of audio files and return a audio embeddings

Take the mean over all non-batch dimensions.

source: https://github.com/openai/guided-diffusion/blob/27c20a8fab9cb472df5d6bdd6c8d11c8f430b924/guided_diffusion/losses.py#L12 Compute the

Get the distribution q(x_t | x_0). :param x_start: the [N x C x ...] tensor of noiseless inputs. :param t: the number of d

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