This project required using RNNs to generate TV show scripts. It was a learning experience, with the most interesting aspect of it being that for an RNN, you have to turn on Retain_Graph for the backpropagation or else you end up with an ever-growing training time as the model tries to adjust weights all the way back over every iteration. This also requires converting the weights to a tuple.
Once again, I learned that you can waste a lot of time on training if the code is wrong. It took me a long time to figure out why I couldn't implement the tuple on the weights (I forgot the square brackets). RNNs are great but after finishing this project I didn't think they were the best way to approach this particular problem.