Enclosed within this folder are the following-

1. Python files - decoder.py, encoder.py, test.py, train.py, utils.py
2. Text files containing losses and scores for each model
3. Text files containing BLEU scores for each sentence in the model. (Structured as {English sentence as the model reads it- includes unknown/start/end tokens}: {score})
4. Graphs plotting the training losses and the differences between the models while testing.

The models reported have been saved onto this google drive link

To run the code, ensure that all python files are in the same directory as well as ensuring that the corpus ted-talks-corpus is in the same directory. Run train.py to run the train the model and run test.py to report scores and save them to a file. All parameters are stored within utils.py and can be adjusted from there.