This repository contains an implementation of the structured attention and compositon model for textual entailment described in the paper Textual Entailment with Structured Attentions and Composition.
To train on the provided sample data and saving, you can simply run:
th trainer.lua --dump model_file
You can find the training parameters and their descriptions in file trainer.lua.
To evaluate the trained model on the dev set, you can run:
th trainer.lua --eval model_file