rt.pos & rt.neg: The dataset contains 10,662 examples of movie review sentences. download from -> "Movie Review data from Rotten Tomatoes". One positive sentences file, one negative sentences file.
vectors.bin: Pre-trained word vectors by google's word2vec, you can download the source-archive from this github project. You can also see the gensim (topic modelling for human) web page for getting more help of how to use this.
data_helpers.py: It contains functions for the data loading (include pre-trained word vectors), data clean and generating batch data for training.
text_cnn.py: The core function for generating a cnn for text classification. Model structure: embedding layer -> convolutional layer -> max-pooling layer -> softmax layer.
train.py: It implements the reading parameters, data preperation and training procedure.