Continual Loader (CLLoader)

A library for PyTorch's loading of datasets in the field of Continual Learning

Aka Continual Learning, Lifelong-Learning, Incremental Learning, etc.

Example:

from torch.utils.data import DataLoader

from clloader import CLLoader
from clloader.datasets import MNIST

clloader = CLLoader(
    MNIST("my/data/path", download=True),
    increment=1,
    initial_increment=5
)

print(f"Number of classes: {clloader.nb_classes}.")
print(f"Number of tasks: {clloader.nb_tasks}.")

for task_id, (train_dataset, test_dataset) in enumerate(clloader):
    train_loader = DataLoader(train_dataset)
    test_loader = DataLoader(test_dataset)

    # Do your cool stuff here

Supported Scenarios

Supported Datasets:

Note that the task sizes are fully customizable.

Furthermore some "Meta"-datasets are available:

• InMemoryDataset: for in-memory numpy array
• PyTorchDataset: for any dataset defined in torchvision
• ImageFolderDataset: for datasets having a tree-like structure, with one folder per class
• Fellowship: to combine several datasets

Sample Images

MNIST:

Task 0	Task 1	Task 2	Task 3	Task 4

FashionMNIST:

Task 0	Task 1	Task 2	Task 3	Task 4

CIFAR10:

Task 0	Task 1	Task 2	Task 3	Task 4

MNIST Fellowship (MNIST + FashionMNIST + KMNIST):

Task 0	Task 1	Task 2

PermutedMNIST:

Task 0	Task 1	Task 2	Task 3	Task 4