This code is a slightly modified implementation of the model CNN-nonstatic mentioned in the paper Convolutional Neural Networks for Sentence Classification (EMNLP 2014). However, other models (e.g, CNN-static) can easily be implemented by tweaking the code in model.py.

There are differences:

• This implementation uses Glove instead of Word2Vec.
• This implementation does not use the L2 norm constraint.

GPU output:

===> Iter 500: | Train acc 0.5755 | Test acc 0.7188
===> Iter 1000: | Train acc 0.7207 | Test acc 0.7348
...
...
===> Iter 5500: | Train acc 0.9816 | Test acc 0.7910