Generative NLP with Variational AutoEncoders

Course details

There’s so much going on in the world of AI right now that it can seem impossible to feel like you’re on track. This is especially true for models that generate text. Which ones should you use, how should you use them, and when?

In this course, join instructor Axel Sirota as he offers an in-depth overview of variational autoencoders (VAEs) specifically focused on how to create complex models in Keras, how to create denoising autoencoders, and how to create VAEs for text generation. Throughout the course, Alex provides hands-on demonstrations and exercise challenges to test out your new skills.

Concepts

0. Introduction

01 - Leveling up in generative NLP
02 - Getting started with VAEs
03 - Getting the most out of this course
04 - Version check

1. Managing Custom Layers, Models, and Functions in Keras

05 - Why do we need custom elements in Keras
06 - Using the functional API vs. sequential API
07 - Demo - Creating a multioutput model
08 - Creating and adding a custom loss function
09 - Demo - Adding a custom loss function
10 - Creating a custom layer
11 - Demo - Training a model with a custom layer
12 - Merging everything - Declaring the model
13 - Demo - Training the custom model
14 - Challenge prep - Create a Siamese network
15 - Solution - Create a Siamese network

2. Autoencoders

16 - What is an autoencoder
17 - Applications of autoencoders
18 - Demo - Creating a simple autoencoder
19 - Using custom encoder to output latent representation
20 - Integrating everything in Keras for text generation
21 - Demo - Creating a deep autoencoder
22 - Challenge - Denoise with an autoencoder
23 - Solution - Denoise with an autoencoder, part 1
24 - Solution - Denoise with an autoencoder, part 2

3. Variational Autoencoders

25 - What changes with VAEs
26 - Incorporating BiLSTMs to VAEs
27 - Demo - Creating the encoder
28 - The loss function and KL divergence
29 - Demo - Creating the custom loss function and custom optimizer
30 - Integrating everything to train the VAE
31 - Generating sentences

4. Challenges

32 - Challenge prep for final challenges
33 - Solution - Building the VAE
34 - Solution - Training the VAE
35 - Solution - Generating text

Conclusion

36 - Recap of what we covered
37 - Where to go next