A small playground for learning how to do data versioning with DVC is established. Version control for data sets tries improve the reproducibility of ML workflows. It allows for more controlled experimentation and development.
DVC can be easily installed through pip install dvc.
Running the script ./make_data.sh creates a directory of example dummy data.
Following that, DVC is intialized by calling dvc init
and the data can be added for tracking by dvc add data/.
pip install dvc
./make_data.sh
dvc init
dvc add data/