add new dementions at the front and the tail

Update stdout by moving cursor up and erasing line for given number of lines. Args: num_lines (int): number of lines

Return batch of w latent codes given a batch of z latent codes. Args: z (torch.Tensor) : Z-space latent code of size [batch_size,

1x1 convolution

1x1 convolution with padding.

3x3 convolution with padding

3x3 convolution with padding.

Crop faces from given images for the given bounding boxes and padding.

Convert time into days, hours, minutes, and seconds string format. Args: t (float): time in seconds Returns (string): "<days

sum over the first and last dimention

Inputs: - img_batch: a torch.Tensor of shape (Batch size, Channels, Height, Width)

Execute an replication callback `__data_parallel_replicate__` on each module created by original replication. The callback will be invoked w

Sample a random latent code from multi-variate standard Gaussian distribution with/without truncation. Args: batch_size (int) : batch s

Save interpretable paths results and create summarizing GIFs. Args: attributes (list) : list of attribute names att

Args: master_callback: a callback to be invoked after having collected messages from slave devices.

Compute the mean and standard-deviation with sum and square-sum. This method also maintains the moving average on the master device.

Create output directory for current experiment under experiments/wip/ and save given the arguments (json) and the given command (bash script).

Create a summarizing GIF image given an images root directory (images generated across a certain latent path) and the number of images to appear a

Create summary .md file for the given attributes group. For all given attributes, show the top-k interpretable paths for each latent code (hash).

Decode locations from predictions using priors to undo the encoding we did for offset regression at train time. Args: loc (tensor):

Detects faces in a given image. This function detects the faces present in a provided BGR(usually) image. The input can be either

Detects faces in a given image. This function detects the faces present in a provided BGR(usually) image. The input can be either

Check if checkpoint file exists (under `self.models_dir`) and set starting iteration at the checkpoint iteration; also load checkpoint weights

Get width and height of images in given list of paths. Images are expected to have the same resolution. Args: img_paths (list): list of i

Initialize model with pretrained weights. Layers that don't match with pretrained layers in name or size are kept unchanged.

Perform L1-normalization.

Log progress in terms of batch accuracy, classification and regression losses and ETA. Args: iteration (int) : current ite

WarpedGANSpace -- Training script. Options: ===[ Pre-trained GAN Generator (G) ]=========================================================

A script for traversing the attribute space of the generated paths (as calculated by `traverse_latent_space.py`). Attribute space includes the fol

An auxiliary script for converting a checkpoint file (`checkpoint.pt`) into a support sets (`support_sets.pt`) and a reconstructor (`reconstructor

Download pre-trained attribute detectors models: -- GAN generators -- Spectral Norm GAN [1] for MNIST [2] and AnimeFaces [3] datas