Applied AI: Building NLP Apps with Hugging Face Transformers

Course details

Looking to expand your skill set in deep learning? Find out how to use Hugging Face transformers to build natural language processing (NLP) applications. In this course, instructor Kumaran Ponnambalam shows you how to build models quickly and easily using pretrained transformers from the Hugging Face library.

Explore models designed for common NLP use cases such as question-answering, text summarization, text generation, translation, and more. Kumaran gives you tips for customizing models with transfer learning to meet the needs of specific use cases—improving your performance and lowering your costs along the way. Develop your know-how to identify and overcome common modeling challenges to ensure successful, error-free deployments. By the end of this course, you’ll be ready to adhere to the best practices and industry standards when you start applying your new skills on the job.

Concepts

0. Introduction

01 - Building NLP apps with Transformers
02 - Course coverage and prerequisites
03 - Setting up the exercise files

1. Question-Answering (Qu-An)

04 - Question-answering in NLP
05 - Types of question-answering
06 - Building a Qu-An pipeline
07 - The SQuAD metric
08 - Evaluating Qu-An performance

2. Text Summarization

09 - Text summarization in NLP
10 - The BART model architecture
11 - Summarization with pipelines
12 - The ROUGE score
13 - Evaluating with ROUGE

3. Natural Language Generation

14 - Natural language generation in NLP
15 - Content creation with Transformers
16 - Conversation generation
17 - Chatbot conversation example
18 - Machine translation in NLP
19 - Translating with Hugging Face Transformers

4. Customizing Models with Transfer Learning

20 - Training a custom model
21 - Loading a Hugging Face dataset
22 - Encoding and preprocessing the dataset
23 - Customizing the model architecture
24 - Training the sentiment model
25 - Predicting with the custom model

5. Deploying and Using Hugging Face Models

26 - Inference challenges with Transformers
27 - Customizing pretrained models
28 - Model compression overview
29 - Serving multiple models

Conclusion

30 - Continuing with Hugging Face