Creates the specified folder if it does not exist. :param path: the complete path of the folder to be created

Channel shuffle: [N,C,H,W] -> [N,g,C/g,H,W] -> [N,C/g,g,H,w] -> [N,C,H,W]

Compute a forward pass. :param x: input tensor (batch_size, input_size) :return: output tensor (10)

Channel shuffle: [N,C,H,W] -> [N,g,C/g,H,W] -> [N,C/g,g,H,w] -> [N,C,H,W]

Compute loss for model. If both `labels` and `mask` are None, it degenerates to SimCLR unsupervised loss: https://arxiv.org/pdf/2002.0

Returns the backbone to be used for to the current dataset.

Creates and returns the training and test loaders for the current task. The current training loader and all test loaders are stored i

Creates and returns the training and test loaders for the current task. The current training loader and all test loaders are stored i

Creates and returns a continual dataset. :param args: the arguments which contains the hyperparameters :return: the continual dataset

Returns the transform used for denormalizing the current dataset.

Returns the transform used for denormalizing the current dataset.

Returns the transform used for normalizing the current dataset.

Returns the transform used for normalizing the current dataset.

Returns the scheduler to be used for to the current dataset.

Returns the transform to be used for to the current dataset.