Configure Your CL Dataset

The continual learning dataset is a sequence of datasets corresponding to continual learning tasks, each of which has their own training and testing data. If you are not familiar with continual learning datasets, feel free to get some knowledge from my CL beginner’s guide.

To configure the continual learning dataset for your experiment, link the cl_dataset field in the main YAML file to sub-config file in the sub-directory /cl_dataset/ of your configs, and specify the class and its arguments in the sub-config. In this package we pre-defined many CL dataset classes in clarena.cl_datasets, see the full list below.

Here is an example:

./clarena/example_configs
β”œβ”€β”€ __init__.py
β”œβ”€β”€ entrance.yaml
β”œβ”€β”€ experiment
β”‚   β”œβ”€β”€ example.yaml
β”‚   └── ...
β”œβ”€β”€ cl_dataset
β”‚   └── permuted_mnist.yaml
...
example_configs/experiment/example.yaml
defaults:
  ...
    - /cl_dataset: permuted_mnist.yaml
  ...
example_configs/cl_dataset/permuted_mnist.yaml
_target_: clarena.cl_datasets.PermutedMNIST
root: data/MNIST
num_tasks: 10
validation_percentage: 0.1
batch_size: 64
num_workers: 10

Supported Dataset List

Here is the full list of supported CL datasets so far. To select a dataset, replace with the class name. Please refer to the documentation of each class to know what fields (arguments) are required for the class.

CL Dataset Class Name Description
Permuted MNIST PermutedMNIST A MNIST variant for CL by random permutation of the input pixels to form differenet tasks
Split MNIST SplitMNIST A MNIST variant for CL by spliting the dataset by class to form different tasks
Permuted CIFAR10 PermutedCIFAR10 A CIFAR-10 permuted variant for CL
Split CIFAR100 SplitCIFAR100 A CIFAR-100 split variant for CL
Back to top