Cerenaut / cfsl

Continual few-shot learning

Continual few-shot learning

Requirements

• PyTorch 1.5.1+
  • Follow instructions here to set it up locally (depends on your environment)

Getting Started

First, you need to setup the CLS module before using it with any of the available frameworks.

1. Change into the cls_module directory
2. Execute the python setup.py develop command to install the package and its dependencies

Frameworks

Omniglot Lake Benchmark

This is an implementation of the one-shot generalization benchmark introduced by Lake. The code is available under the directory frameworks/lake.

To run an experiment using the Lake framework, you will need a valid configuration file. There is an existing configuration file located in frameworks/lake/config.json with the default configuration.

Run the experiment using python oneshot_cls.py --config path/to/config.json

CFSL Benchmark

The code is available under frameworks/cfsl and is derived from https://github.com/AntreasAntoniou/FewShotContinualLearning

To run the experiments with CLS, you can simply modify the configuration file in omniglot_cls.json and then run the experiment using bash omniglot_cls.sh GPU_ID latest.

Note: Set GPU_ID to 0 if you are not using a GPU, and 1 if you are using a GPU.