This repository contains code of Jung et al's ACL 2020 paper titled "Posterior Calibrated Training on Sentence Classification Tasks"
We use python 3.7. Please run pip install -r requirement.txt
to install python dependencies.
Note that you add the flag --poscal_train
for PosCal training. The example below uses the ShortRomance datsaet in xSLUE (Kang et al., 2019). Please refer to the xslue resposiotry to download the datasets.
python classify_bert.py \
--model_type bert \
--model_name_or_path bert-base-uncased \
--task_name ShortRomance \
--do_train \
--poscal_train \
--data_dir $PATH/to/the/data \
--output_dir $PATH/to/the/output
@inproceedings{jung20acl_poscal,
title = {Posterior Calibrated Training on Sentence Classification Tasks},
author = {Taehee Jung, Dongyeop Kang, Hua Cheng, Lucas Mentch, and Thomas Schaaf},
booktitle = {2020 Annual Conference of the Association for Computational Linguistics (ACL)},
url = {https://arxiv.org/abs/2004.14500},
year = {2020}
}