Welcome to IIRC’s documentation!

iirc is a package for adapting the different datasets (currently supports CIFAR-100 and ImageNet) to the iirc setup and the class incremental learning setup, and loading them in a standardized manner.

lifelong_methods is a package that standardizes the different stages any lifelong learning method passes by, hence it provides a faster way for implementing new ideas and embedding them in the same training code as other baselines, it provides as well the implementation of some of these baselines.

Project Homepage | Project Paper | Source Code | PyPI Package

Installation:

you can install the iirc package using the following command

pip install iirc

To use it with PyTorch, you will need as well to install PyTorch (1.5.0) and torchvision (0.6.0)

Guide:

• IIRC Package Tutorial Using PyTorch
• Loading IIRC and incremental datasets
• Lifelong Learning Methods Guide

Documentation:

• iirc package
• lifelong_methods package

• Index

• Module Index

• Search Page

Paper

IIRC is introduced in the paper “IIRC: Incremental Implicitly-Refined Classification ”. If you find this work useful for your research, this is the way to cite it:

@misc{abdelsalam2021iirc,
    title = {IIRC: Incremental Implicitly-Refined Classification},
    author={Mohamed Abdelsalam and Mojtaba Faramarzi and Shagun Sodhani and Sarath Chandar},
    year={2021}, eprint={2012.12477}, archivePrefix={arXiv},
    primaryClass={cs.CV}
}

Community

If you think you can help us make the iirc and lifelong_methods packages more useful for the lifelong learning community, please don’t hesitate to submit an issue or send a pull request.