Readme

###############################
Name:Ashwin Nair Anilil
################################

Run the evaluate_xtest.py to get the prediction for xtest.txt

Run the train_with_xtrain_and_ytrain.py to train the model parameters can be changed inside.
The parameters used now are from the final model used for submission.

############################################################

The approach used is heavealy borrowed from the papers https://arxiv.org/abs/1408.5882 and https://arxiv.org/abs/1510.03820.

The code snippets are borrowed from https://github.com/jiegzhan/multi-class-text-classification-cnn and https://github.com/dennybritz/cnn-text-classification-tf

I had decided to use this approach because I have previously worked on CNNs and LSTMs with tensorflow for my thesis.

The network is a bit overfit looking at the tersorflow loss graphs which can be accessed with the tensorboard.