A tutorial that I had put together for explaining LSTMs and their advantages compared to RNNs and presented recently The objective of the tutorial was to come out with an understanding of:
- What are RNNs and what are the applications of RNNs?
- Why choose LSTMs over RNNS?
- Where to apply LSTMs?
The presentation is followed by a code walkthrough of using LSTMs for a Email Generation use case where a sequence to sequence LSTM model is used. The model is trained on text from conversations (email) and generate.
Note: This is a skeletal repository of my work and does not contain data that was originally used for the model due to privacy reasons.