Expand the tensor `v` to the dim `dims`. Args: `v`: a PyTorch tensor with shape [N]. `dim`: a `int`. Returns: a P

Load model from file. Args: path (str): file path

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

Initialize a pretrained Real-ESRGAN upsampler. Args: model_name_or_path (str, optional): The Hugging Face repo ID or path to local

Load the different models. We also reuse the models that are already in memory to speed things up instead of loading them again.

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

Compute sigma_t of a given continuous-time label t in [0, T].

Converts a denoiser output to a Karras ODE derivative.

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

Sets metadata on image, in both text form and as an ImageMetadata object

Convert the model to the noise prediction model or the data prediction model.

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

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

Simple function to allows us to change the title dynamically. Normally you can use `st.set_page_config` to change the title but it can only b

Simple function to allows us to change the title dynamically. Normally you can use `st.set_page_config` to change the title but it can only b

Determines and returns the next sequence number to use when saving an image in the specified directory. If a prefix is given, only consi

saves the image according to selected parameters. Expects to find generation parameters on image, set by ImageMetadata.set_on_image()

Appends dimensions to the end of a tensor until it has target_dims dimensions.

Same as diffusers `from_pretrained` but with tiled option, which makes images tilable

r""" Enable sliced attention computation. When this option is enabled, the attention module will split the input tensor in slices, to

Gets metadata from an image, first looking for an ImageMetadata, then if not found tries to construct one from the info

Create a linear module.

Load the different models. We also reuse the models that are already in memory to speed things up instead of loading them again.

Helper to get the SessionInfo and JobInfo.

Performs all_gather operation on the provided tensors. *** Warning ***: torch.distributed.all_gather has no gradient.