Dataset and code for “Going on a vacation” takes longer than “Going for a walk”: A Study of Temporal Commonsense Understanding EMNLP 2019. (link)

We provide the dev/test split as specified in the paper, along with a detailed readme.txt file under dataset/

See the details and instructions at: http://leaderboard.allenai.org/mctaco

At this point, we provide the outputs of the ESIM/BERT baselines.

To run BERT baseline:

First install required packages with:

pip install -r experiments/bert/requirements.txt

Your the following command to reproduce BERT predictions under ./bert_output:

sh experiments/bert/run_bert_baseline.sh

Evaluate the predictions with which you can further evaluate with the following command:

python evaluator/evaluator.py eval --test_file dataset/test_9442.tsv --prediction_file bert_output/eval_outputs.txt

ESIM baseline: Releasing soon after some polish

See the following paper:

@inproceedings{ZKNR19,
    author = {Ben Zhou, Daniel Khashabi, Qiang Ning and Dan Roth},
    title = {“Going on a vacation” takes longer than “Going for a walk”: A Study of Temporal Commonsense Understanding },
    booktitle = {EMNLP},
    year = {2019},
}