fast.ai
Offers deep learning courses, the fastai library, and a companion book, helping programmers train models with little code and learn deep learning systematically.
About fast.ai
fast.ai is a deep learning education and research website comprising three parts: courses (such as Practical Deep Learning for Coders and the new course How To Solve It With Code announced in 2024), software (the fastai deep learning library and nbdev), and the companion book Practical Deep Learning for Coders with fastai and PyTorch. The fastai library provides practitioners with high-level components that make it possible to quickly achieve state-of-the-art results in standard deep learning domains with very little code; it also gives researchers freely composable low-level components for building new methods. The website also publishes blog posts on deep learning and education.
Best for
Best use cases
Want to learn deep learning systematically through courses
Quickly train image, text, tabular, or recommendation system models with a small amount of code
Mix and match low-level components to build new methods in research
Features
Layered API design
Provides high-level components that let you quickly achieve state-of-the-art results in standard deep learning domains with very little code, as well as composable low-level components for researchers to build new methods.
Multi-domain applications
Official tutorials show how to build image classification, image segmentation, text sentiment, recommendation system, and tabular data models with about 5 lines of code, with largely identical code across application types.
Get started without installation
Can be used without installation via Google Colab, and every page of the documentation can be opened as an interactive notebook; it can also be installed locally with pip install fastai.
Courses and books
Offers courses such as Practical Deep Learning for Coders, the companion book, and the new course How To Solve It With Code announced in 2024.