Chameleon

Introduction

In this repository you will find a pytorch implementation of Chameleon with the short-term and long-term stores.

Project Structure

The project is structured as follows:

• models/: In this folder the main MobileNetV1 model is defined along with the custom Batch Renormalization Pytorch layer.
• chameleon.py: Main Chameleon algorithm.
• data_loader.py: CORe50 data loader.
• params.cfg: Hyperparameters that will be used in the main experiment on CORe50-NI.
• README.md: This instructions file.
• utils.py: Utility functions used in the rest of the code.

Getting Started

When using anaconda virtual environment all you need to do is run the following command and conda will install everything for you. See environment.yml:

conda env create --file environment.yml
conda activate chameleon-env

Then to reproduce the results on the CORe50-NI benchmark you just need to run the following code:

python chameleon.py --run 3 --scenario ni