Pytorch--Text matching

This is a code for text matching,all the Deep model is run on the platform of pytorch

It is a competition about CIKM spanish text matching

Environment

python 3.6 or python 3.5
pytorch 0.4.0 for cpu or gpu (This code main runs on cpu)
tqdm
scikit-learn

Code organization

data is used to save train,test,word-embeding vector or temporary file
model is used to save all kinds of Deep models
stacking is used to save predicted result on validation set of all kinds of models
data_propresse is used to preprocess data
submit is used to save submit file

Data preprocess

run the file runme.py in the dir data_propresse

Train

run the file train.py

ensemble

run the file ensemble2.py