Meta-learning codebase

Env setup

In order to use the code build the docker from the repository folder:

docker build -t meta_learning:v1.0 -f docker/Dockerfile .

or make the conda environment:

conda env create -f docker/conf_files/metalearning_pytorch.yml

Dataset setup

In order to use docker the dataset paths should be set as env variables (either manually or in docker/run_docker.sh) then run run_docker.sh.

The folder structure should be the following (running run_docker.sh will automatically make this structure according to docker-compose.yml):

├── parent_folder
    ├── meta_learning
        ├── {the code ...}      
    ├── data
        ├── imagenet
        ├── imagenet64
        ├── {other datasets ...}
    ├── results
        ├── {experiment result folders will be saved here}