This folder contains a dictionary of optimal parameters for each recommendation environment. The settings in best_params.json were found via hyper-parameter tuning and were used in our experiments. Recall that each environment consists of a dataset and a recommender.

Upon first training a recommender for a dataset the default behaviour is to save a .pkl file containing the trained recommender model. This makes subsequent instantiations of Environments much faster.