SEQ CIFAR10 224#

Classes#

class datasets.seq_cifar10_224.SequentialCIFAR10224(args)[source]#

Bases: ContinualDataset

Sequential CIFAR10 Dataset. The images are resized to 224x224. Version with ViT backbone.

Parameters:
  • NAME (str) – name of the dataset.

  • SETTING (str) – setting of the dataset.

  • N_CLASSES_PER_TASK (int) – number of classes per task.

  • N_TASKS (int) – number of tasks.

  • N_CLASSES (int) – number of classes.

  • SIZE (tuple) – size of the images.

  • MEAN (tuple) – mean of the dataset.

  • STD (tuple) – standard deviation of the dataset.

  • TRANSFORM (torchvision.transforms) – transformations to apply to the dataset.

MEAN = (0.4914, 0.4822, 0.4465)#
NAME: str = 'seq-cifar10-224'#
N_CLASSES: int = 10#
N_CLASSES_PER_TASK: int = 2#
N_TASKS: int = 5#
SETTING: str = 'class-il'#
SIZE: Tuple[int] = (224, 224)#
STD = (0.247, 0.2435, 0.2615)#
TEST_TRANSFORM = Compose(     Resize(size=224, interpolation=bilinear, max_size=None, antialias=True)     ToTensor()     Normalize(mean=(0.4914, 0.4822, 0.4465), std=(0.247, 0.2435, 0.2615)) )#
TRANSFORM = Compose(     Resize(size=224, interpolation=bilinear, max_size=None, antialias=True)     RandomCrop(size=(224, 224), padding=28)     RandomHorizontalFlip(p=0.5)     ToTensor()     Normalize(mean=(0.4914, 0.4822, 0.4465), std=(0.247, 0.2435, 0.2615)) )#
get_backbone()[source]#
get_batch_size()[source]#
get_class_names()[source]#
get_data_loaders()[source]#

Class method that returns the train and test loaders.

Return type:

Tuple[DataLoader, DataLoader]

static get_denormalization_transform()[source]#
get_epochs()[source]#
static get_loss()[source]#
static get_normalization_transform()[source]#
static get_prompt_templates()[source]#
static get_transform()[source]#