This Project was done by Prateek Yadav under the supervision of professor Ambedkar Dukkipati as a part of the graduate level course E0_250 Deep Learning at his lab Statistics and Machine Learning Group at Indian Institute of Science, Bangalore. The code is in LUA using pakage torch7 and cudatorch. so you'll need a GPU.
In this project my task was to work on several data sets including MR(Movie Review), SST-1, SST-2, SUBJ, TREC, CR, MPQA try out different settings and various values of hyper parameters and to find out what would be the best setting and compare it with the results obtained by Yoon Kim In his paper "Convolutional Neural Networks for Sentence Classification" and to include various optimization algorithms to improve upon the result. There were 4 model as in the paper and all those were tested and tried out. Slight vartions in architecture was also tried out.
This github Repo only contains code and datasets for the file which contains the trained model and all other material are also availabe at openly here