Move a recipe and all its movable registered objects to a device Args: device: a torch device

Initialize a module with kaiming normal init Args: m (nn.Module): the module to init a (float): the slope of the nonlinearit

Iterate over all nn.Modules registered in the recipe

Return the value averaged over the window

Register an object into the recipe as a member. Calling :code:`recipe.register('foo', bar)` registers bar, and makes it usable

Run the recipe for :code:`epochs` epochs. Args: epochs (int): number of epochs Returns: The state

Move a recipe and all its movable registered objects to cpu

Move a recipe and all its movable registered objects to cuda

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

A 1x1 Conv2d

Log `x`

Remove BatchNorm in Sequentials / CondSeqs in a smart way, restoring biases in the preceding layer.

r"""Performs a single optimization step. Arguments: closure: A closure that reevaluates the model and returns the loss.

A 3x3 Conv2d with 'same' padding

Gradient penalty, mainly for GANs. Of the form :code:`E[(||dmodel(data)/ddata|| - objective_goal)²]` Args: model (function / nn.

Restore a recipe

Sample a value.

Initialize a module with xavier normal init Args: m (nn.Module): the module to init dynamic (bool): wether to scale the weig

Sample a batch of images Args: temp (float): the sampling temperature N (int): number of images to generate

Indent :code:`text` by :code:`amount` spaces. Args: text (str): some text amount (int): an indentation amount Returns:

Return the tensor on cpu and detached, ready to be transformed to a PIL image

Replace BatchNorm with InstanceNorm.

0-GP from Improving Generalization And Stability Of Generative Adversarial Networks ( https://arxiv.org/abs/1902.03984 ). Args:

A Conv2d with 'same' padding

Extends Classification with default cross entropy forward passes. Also adds AdamW/SGD and CosineDecay schedule Inherited training callba

Two nested loops, usually one for training and one for testing, but can serve other purposes. The model is automatically registered and c

Args: lr (float, optional): the learning rate visdom_env (str or None): the name of the visdom env to use, or None

Average pool to spatial size :code:`size` rather than 1. Recreate the first Linear to accomodate the change.

Get the HTML inspector view.

Remove pooling and flattening operations, convert linears to conv1x1

Load an ImageFolder dataset faster by caching the file list the first time it is accessed. Force refreshing by deleting :code:`{root}/cac

Train a model and evaluate it with a custom function. The model is automatically registered and checkpointed as :code:`checkpoint['model']`,

Return the batched Gram matrix of `m` Args: m (tensor): a matrix of dim 3, first one is the batch Returns: The batch of

r""" Compute the cross entropy between the logits `pred` and a normalized distribution `soft_targets`. If `soft_targets` is a one-hot vector,

Recursively index a `dict` by a hierarchical key ``` >>> dict_by_key({'a': [{'b': 42}]}, 'a.0.b') 42 ``` Args: d (d

Freeze all parameters of `net`