LSTM_Text_Generation

This is a project from my time at General Assembly. contains code for text generation through Keras.

Be advised, this is a work in progress

This repo contains the jupyter notebook with code for

1. Cleaning your corpus (from a dataframe)
2. Preprocessing steps
3. Model setup
4. Text generation setup

Text Generator

If you find this repo and the code has not been highlighted with its purpose in the process, I will get to it shortly. This is definitely a learned experience, so add things as you see fit.

Project Details

I trained an LSTM on a corpus of speeches from 6 notable dictators. My hope was to generate convincing text in the style of these dictators. Within the repository, you will find a csv containing portions of the corpus with label indicating the speaker. I would encourage adding to this corpus. As far as I know, there aren't any labeled sources out there. This could also he helpful for a classification problem.

Credits

This code uses insights from Jason Brownlee, Nathan Raw, J Beightol and James Babyak