Apply the block to a Tensor, conditioned on a timestep embedding. :param x: an [N x C x ...] Tensor of features. :param em

r"""x: [B, C, H, W]. context: [B, L, C] or None.

r"""x: [B, L, C]. mask: [*, L, L].

# modified version of https://github.com/assafshocher/BlindSR_dataset_generator # Kai Zhang # min_var = 0.175 * sf # variance of the

Returns: The rank of the current process within the local (per-machine) process group.

Returns: The size of the per-machine process group, i.e. the number of processes per machine.

Initialize variables needed for distributed training.

Initializes the default process group. Args: local_rank (int): the rank on the current local machine. local_world_size (int):

r"""DDPM sampling loop for inpainting, with Manifold Constrained Gradient (MCG) correction.

r"""DDPM sampling loop for inpainting.

Inverse the sample `x` from time `t_start` to `t_end` by DPM-Solver. For discrete-time DPMs, we use `t_start=1/N`, where `N` is the t

r"""Inverse map of indices. E.g., if A[indices] == B, then B[inv_indices] == A.

Determines if the current process is the master process.

Logs json stats. Args: stats (dict): a dictionary of statistical information to log.

Args: img: numpy image, WxH or WxHxC sf: scale factor Return: cropped image

The noise predicition model function that is used for DPM-Solver.

r"""Create a wrapper function for the noise prediction model. DPM-Solver needs to solve the continuous-time diffusion ODEs. For DPMs trained on d

r"""Sample from p(x_{t-1} | x_t) p(x_{t-2} | x_{t-1}) ... p(x_0 | x_1).

r"""Sample from p(x_{t-1} | x_t) p(x_{t-2} | x_{t-1}) ... p(x_0 | x_1).