Computes and stores the average and current value

ImageNet Dataset. Arguments: - root_dir (:obj:`str`): root directory of dataset - meta_file (:obj:`str`): name of meta

Subclass torch's LayerNorm to handle fp16.

Nearest Neighbour Memory Bank implementation This class implements a nearest neighbour memory bank as described in the NNCLR paper[0]. Durin

Arguments: pipeline (Pipeline): a :class:`linklink.dali.CustomPipeline` which will be running. batch_size (int):

Gaussian blur augmentation in SimCLR https://arxiv.org/abs/2002.05709

Subclass torch's LayerNorm to handle fp16.

A ResNet class that is similar to torchvision's but contains the following changes: - There are now 3 "stem" convolutions as opposed to 1, wi

A ResNet class that is similar to torchvision's but contains the following changes: - There are now 3 "stem" convolutions as opposed to 1, wi

Hide common options away.

r"""Implements layer-wise adaptive rate scaling for SGD, based on `"Large Batch Training of Convolutional Networks" <https://arxiv.org/abs/1708.03

Randomly choose one of the best 24 Sub-policies on ImageNet. Example: >>> policy = ImageNetPolicy() >>> transformed = policy

r"""The normalized temperature-scaled cross entropy loss, based on `"A Simple Framework for Contrastive Learning of Visual Representations" <https

r"""The normalized temperature-scaled cross entropy loss, based on `"A Simple Framework for Contrastive Learning of Visual Representations" <https

First resize a image to SIZE in the minimum side. Then conduct random crop

Take two random crops of one image as the query and key.

Take two random crops of one image as the query and key.

Perform gamma correction on an image.

Randomly choose one of the best 25 Sub-policies on CIFAR10. Example: >>> policy = CIFAR10Policy() >>> transformed = policy(i

Clip Dataset. Arguments: - root_dir (:obj:`str`): root directory of dataset - meta_file (:obj:`str`): name of meta file

r"""Set the learning rate of each parameter group using a cosine annealing schedule. Arguments: - optimizer (:obj:`Optimizer`): W

r"""CustomPipeline will work with :class:`linklink.dali.DataLoader` to provide pytorch native dataloader experience.

Randomly mask out one or more patches from an image.

Randomly mask out one or more patches from an image. Args: n_holes (int): Number of patches to cut out of each image. length (int)

r"""Implements RMSprop algorithm in FP16. Proposed by G. Hinton in his `course <http://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slide

r"""Implements stochastic gradient descent (optionally with momentum) in FP16. Nesterov momentum is based on the formula from `On the importa