Large-scale Cloze Test Dataset Designed by Teachers

Code for the paper:

Large-scale Cloze Test Dataset Designed by Teachers. Qizhe Xie*, Guokun Lai*, Zihang Dai and Eduard Hovy.

An example problem:

• Correct answers are highlighted

Dependencies

• Python 2.7
• latest Pytorch

Usage

• tar xzvf data.tar.gz
• test the best model: bash test.sh
• replace embed/glove.6B.300d.txt with the corresponding file in http://nlp.stanford.edu/data/glove.6B.zip
• training the best model: bash train.sh

Acknowledgement

• The code is adapted from OpenNMT (https://github.com/OpenNMT/OpenNMT-py) and Pytorch language model example (https://github.com/pytorch/examples/tree/master/word_language_model)
• The code to highlight informative words (Figure 1 in the paper) is available at https://gist.github.com/ihsgnef/f13c35cd46624c8f458a4d23589ac768

Contact

• Please contact Qizhe Xie (qizhex AT cs cmu edu) if you have questions

License

MIT