Applied AI: Getting Started with Hugging Face Transformers

Course details

Using pretrained transformers for natural language processing (NLP) has become extremely popular among ML engineers and data scientists. If you work in the field, or even have a role adjacent to it, you need to stay apace with the latest innovative tools. In this course, instructor Kumaran Ponnambalam shows you how to take your AI skills to the next level using the repository of pretrained transformers available on the Hugging Face platform.

Explore strategies to leverage transformers for various use cases to solve complex problems and deliver consistent, timely results. Kumaran shows you how to implement two simple NLP use cases—sentiment analysis and named entity recognition—utilizing pretrained transformers from Hugging Face and customizing them as you go. Upon completing this course, you’ll be prepared to apply your new AI skills to projects in your current or future role.

Concepts

0. Introduction

01 - Getting started with Transformers
02 - Course coverage and prerequisites
03 - Setting up the exercise files

1. Machine Learning for NLP

04 - Natural language processing
05 - ML process for NLP
06 - Labeling for NLP
07 - Tokenization
08 - Vectorization

2. Introduction to Transformers

09 - What is a Transformer
10 - Positional encoding
11 - Attention in Transformers
12 - The encoder
13 - The decoder
14 - Transformer training and inference

3. Pretrained Transformers

15 - Challenges with building Transformers
16 - A language model
17 - Pretrained Transformer models
18 - The BERT Transformer
19 - The GPT Transformer
20 - The T5 Transformer

4. Introduction to Hugging Face

21 - Introduction to Hugging Face
22 - Pretrained models in Hugging Face
23 - Datasets in Hugging Face
24 - Pipelines in Hugging Face
25 - Training with Hugging Face

5. Sentiment Analysis with Hugging Face

26 - The sentiment analysis problem
27 - Reviewing pipeline tasks
28 - Loading a pipeline
29 - Predicting sentiment with Pipelines
30 - Using a custom model

6. Named Entity Recognition

31 - Introduction to named entity recognition
32 - Running the standard NER pipeline
33 - Understanding the model architecture
34 - Reviewing model configuration
35 - Using a custom model and tokenizer

Conclusion

36 - Continuing with Transformers