Compute the entropy of the categorical distribution specified by the logits `out` along dimension `dim`. Args: out (tensor): log

Set :code:`.grad` to None for all parameters instead of zeroing out. It is faster.

Return the tensor

Get parameters for ``crop``. Args: img (PIL Image): Image to be cropped. scale (float): range of size of the origin si

Return the Gram matrix of `m` Args: m (tensor): a matrix of dim 2 Returns: The Gram matrix

r""" Inverse or lerp. For `t` between `a` and `b`, returns the fraction or `a` and `b` in `t`. :math:`\frac{t - a}{b - a}` Args:

Convert an image to a b64 html inline image. Possible options: width (int, pixels), height (int, pixels) or jpgquality (int, percentage)

Insert module :code:`new` with name :code:`name` after element :code:`key` in sequential :code:`base` and return the new sequence.

Decorrelate the color of the image `t` and return the result

Logs hyper parameters and results.

Change all relus into leaky relus for modules and submodules of net.

r""" Mixes samples `x1` and `x2` with respective labels `y1` and `y2` according to MixUp :math:`\lambda \sim \text{Beta}(\alpha, \alpha)`

Counts the number of parameters of `net` Args: net (nn.Module): the net Returns: the number of params

Initialize a module with gaussian weights of standard deviation std Args: m (nn.Module): the module to init Returns: th

Patch Discriminator from pix2pix

The architecture used in `Pix2Pix <https://arxiv.org/abs/1611.07004>`_, able to train on 256x256 or 512x512 images.

Recursively call state_dict() on all elements contained in a list / tuple / dict so that it can be saved safely via torch.save(). Args:

Remove a weight_scale hook previously applied on :code:`getattr(module, name)`.

Restore BatchNorm if deleted

Randomly scale the image horizontally by a factor [1 - x; 1 + x] and vertically by a factor of [1 - y; 1 + y]. Args:

Send all tensors contained in `x` to `device`, when `x` is an arbitrary nested datastructure of dicts and lists containing tensors Args:

Returns: A state dict

Update the weights Args: closure (optional fn): a function that computes gradients

Update the weights Args: closure (optional fn): a function that computes gradients

Performs a single optimization step. Arguments: closure (callable, optional): A closure that reevaluates the model

Performs a single optimization step. Arguments: closure (callable, optional): A closure that reevaluates the model

Performs a single optimization step. Args: closure (callable, optional): A closure that reevaluates the model and

Step the scheduler to another iteration

Tempered softmax. Computes softmax along dimension 1 Args: x (tensor): activations t (float): temperature n_iters (i

Set the classifier architecture to avgpool-flatten-linear.

Transform the convolution into a hopefully equivalent transposed convolution

Set the classifier architecture to avgpool-flatten-linear1-relu-dropout-linear2-relu-dropout-linear3, like initially done wit