evaluate.py
is a utility to evaluate your EmotionX 2019 submission file and calculate the model's performance.
To use, simply run:
% python3 evaluate.py --goldfile goldfile.json --predfile predfile.json
where
goldfile.json
is the gold-labeled evaluation file
predfile.json
is your submission file (with predicted labels)
You will need to run it once for every dataset. For example:
% python3 evaluate.py --goldfile friends_eval_gold.json --predfile friends_pred.json
% python3 evaluate.py --goldfile emotionpush_eval_gold.json --predfile emotionpush_pred.json
verify.py
is a utility to verify your EmotionX 2019 submission file and make sure it's in the correct format.
To use, simply run:
% python3 verify.py --testfile testfile.json --predfile predfile.json
where
testfile.json
is the unlabeled evaluation file
predfile.json
is your submission file (with predicted labels)
For example, for the Friends dataset, use:
% python3 verify.py --testfile friends_eval.json --predfile friends_pred.json
If Python complains about a missing module, install it using pip (or pip3).