Baseline classifier model using wandb experiment tracker and pytorch lightning framework. Allows reduction of dataset and splitting into "unlabelled" and labelled datasets for easy comparison with semi-supervised learning.