To recreate our results in table 5: 1. Install requirements with 'conda env create -f environment.yml' 2. Activate the environment: 'conda activate amkg' 3. Add the data csv files to the data directory 4. Train a model with the required configuration: 'python train.py -datadir ./data/{dataset name} -knowledge {knowledge source} -save {save directory} -model {model architecture} -bsz 10 -lr 0.00001 -epochs 10' For datasets you may choose 'student_essay.csv' or 'debate.csv'. Knowledge sources are: none, conceptnet, atomic2020, comet and comet_link Model architectures: hybrid, attn (hybrid+) and crossencoder Crossencoder is only used when knowledge source is 'none'. To train the link model first download the data at https://allenai.org/data/atomic-2020 into data/atomic2020 and then run: 'python link.py -save {save directory} -bsz 16 -lr 0.00001 -epochs 5'