This is the repo for AI CUP 2019: Abstract Labeling and Classification (Discussion forum)
We share our report that displays the details of our training process.
pip install -r requirements.txt
(sudo) python -m spacy download en_core_web_sm
Use above command to install required packages
python csv2json.py [trainset_path] 'train'
python csv2json.py [public_testset_path] 'test'
python csv2json.py [private_testset_path] 'test'
Use the command python csv2json.py [datapath] [mode] to preprocess training and testing data
bash scripts/train.sh output/all
bash scripts/train_BACKGROUND.sh output/background
bash scripts/train_OBJECTIVES.sh output/objectives
...
Update the scripts/train.sh script with the appropriate datapaths.
mkdir preds
bash scripts/predict_public.sh
bash scripts/predict_private.sh
Use above command to predict labels, then use python submission.py to get the final prediction file final_submission.csv