-
The dataset and splits are available in the ESI, please download them.
-
Modify the sample_config.yaml to set the datasets paths to raw_test.pickle, raw_train.pickle, raw_val.pickle
-
Create a conda environment using the requirements.yaml, or using pip
python run.py /PATH/TO/CONFIG.yaml
This will run the train test script and log the resutls to wandb.