Compute lambda_t = log(alpha_t) - log(sigma_t) of a given continuous-time label t in [0, T].

Kernel for computing the matmul C = A x B. A has shape (M, K), B has shape (K, N) and C has shape (M, N)

Generate samples from the model. :param model: the model module. :param shape: the shape of the samples, (N, C, H, W).

Kernel for computing the matmul C = A x B. A has shape (M, K), B has shape (K, N) and C has shape (M, N)

Parse unknown args.

Perform a download for a pipe: url.

3D pixelshuffle operation.

3D pixel unshuffle operation.

A few custom initialization steps that should be called after the object is created. Currently, the only one we have is to bind a few

A few custom initialization steps that should be called after the object is created. Currently, the only one we have is to bind a few

Get previous latent tail.

Kernel for computing the matmul C = A x B. A has shape (M, K), B has shape (K, N) and C has shape (M, N)

Generates a random 3x3 rotation matrix using a random unit quaternion. This provides a uniform distribution of rotations.

check if can generate more

Args: input_dict (dict): all the values will be reduced average (bool): whether to do average or sum Reduce the values in the

Reset all interface state

Do spatial cropping and resizing to the video clip Args: clip (torch.tensor): Video clip to be cropped. Size is (T, C, H, W)

Run the inpainting pipeline with the current image, mask, and prompt. This is where you can implement your inpainting model!

Importance-sample timesteps for a batch. :param batch_size: the number of timesteps. :param device: the torch device to save

Compute the sample at time `t_end` by DPM-Solver, given the initial `x` at time `t_start`. =========================================

Forward-only monotonic sequences (non-decreasing in t across frames). Returns [B, F] int64 tensor.