Free course on Natural Language Processing (NLP) from The Hugging Face

The Hugging Face, a data science platform with a community of scientists, researchers, and ML engineers who contribute to open source projects, offers a free course to take you from novice to expert in natural language processing.

Natural Language Processing or NLP is a field of artificial intelligence that makes computers understand natural languages ​​like English. And what? Why invest in learning NLP at all?

NLP tries to find meaning in textual data, which is much more difficult than doing the same with numeric data. NLP is ubiquitous because people communicate in language for just about everything: Internet searches, advertising, email, customer service, language translation, virtual agents, medical reports, and so on. Many organizations are looking to integrate NLP into their workflows and the products they provide, such as translation, speech recognition, and chatbots. Sounds like a good career move.

In this course, you will be able to learn about natural language processing using libraries from the Hugging Face ecosystem.

The course begins with an introduction to the Transformers libraries in chapters 1-4, which discuss how models work. This includes looking at the Encoder, Decoder, and Sequence-to-sequence models, fine-tuning these models with the Trainer API or Keras, and an introduction to The Hugging Face Hub’s pre-trained models.

Chapters 5-7 cover the basics of working with datasets and tokenizers before diving into classic NLP tasks. This includes building your own dataset, doing semantic search with FAISS, learning a new tokenizer from an old one, and building the tokenizer block by block.

Chapter 8 includes instructions for debugging errors and asking for help if needed.

The final Chapter 9 shows you how to create interactive demos for your machine learning models.

The prerequisite barrier is low – you should be comfortable with Python and have some knowledge of high school math. No previous knowledge of NLP or machine learning is assumed, but some familiarity with PyTorch or TensorFlow is preferred.

In terms of time, despite the fact that the course is designed for self-study, each chapter can be completed in 1 week if you give it 6-8 hours a week.

The course can be studied on the official website, but it can also be studied as a Youtube playlist with a length of 79 videos. True, the official website has links to the code that can be found in each section that can be run in Google Colab or Amazon SageMaker Studio Lab.