Final project for COMP 6630 Machine Learning. This class was taught by Professor Bo Liu at Auburn University.
Link for GloVe dataset
https://nlp.stanford.edu/projects/glove/
- /data - contains data that was used to train all of the models, as well as the Python script used to scrape the data. The data is organized by genere.
- /gan - contains source code for the GAN model, along with saved model states of each nth training iteration and generated song results
- /lstmGAN - contains source code for the GAN model with the LSTM generator, along with saved model states of each nth training teration and generated song results
- /lyric_outputs - contains genreated lyrics for the LSTM baseline model with and without using the GloVe dataset enhancer
- /provided - documents provided for the assignment from the class
- .gitignore - git file for ignoring certain local files when pushing changes
- LICENSE - license for using this project
- README.md - this file that you are reading
- gan-generator-discriminator-loss.png - training loss results image for the basic GAN model
- lstm-medium-glove-train-accuracy.png - training accuracy results image for the LSTM model using the medium GloVe dataset
- lstm-medium-glove-train-loss.png - training loss results image for the LSTM model using the medium GloVe dataset
- lstm-none-glove-train-accuracy.png - training accuracy results image for the LSTM model using no GloVe dataset
- lstm-none-glove-train-loss.png - training loss results image for the LSTM model using no GloVe dataset
- lstm.py - source code for the LSTM baseline model
- requirements.txt - Python dependencies for this project