REPLICATING.md

To replicate the experiments described in the paper,

1. Install requirements:

pip install -r requirements.txt

2. Get, convert and split the data:

cd data
make split

3. Train the models:

slurm bash/train_saga.sh bert
slurm bash/train_saga.sh bert_feats
slurm bash/train_saga.sh feats
slurm bash/train_saga.sh glove
slurm bash/train_saga.sh glove_feats

4. Evaluate the models:

slurm bash/predict_saga.sh bert
slurm bash/predict_saga.sh bert_feats
slurm bash/predict_saga.sh feats
slurm bash/predict_saga.sh glove
slurm bash/predict_saga.sh glove_feats