Install conda environment from env.yaml. If the survae module cannot be found, clone the repo and install with pip.

Sweeps over seeds and hyperparameters were performed using the config yaml files found in sweeps and train_sweep.py.

Single runs can be run with python train.py and make use of default values provided in the yaml files in configs.