Home

Awesome

SciTail

This released code is for our ACL2018 paper "End-Task Oriented Textual Entailment via Deep Explorations of Inter-Sentence Interactions". It gets STOA performance in a textual entailment benchmark -- 82.1% accuracy on SciTail dataset. We release the code and the pretrained model

To reproduce the result, just run the "train_SciTail_DeIsTe_model.py" file. It needs word2vec embeddings. We provide the pretrained model so that reloading it can reproduce the paper numbers.