D2L

Summary

This is the documentation for the interactive deep learning book 'Dive into Deep Learning', implemented with several different frameworks and adopted by 500 universities in 70 countries.

Abstract

'Dive into Deep Learning' is an interactive deep learning book that provides code, math, and discussions. It is implemented with PyTorch, NumPy/MXNet, JAX, and TensorFlow, and is adopted at 500 universities from 70 countries. The book covers a wide range of topics in deep learning and is forthcoming on Cambridge University Press. It is also available in Chinese and can be run for free on SageMaker Studio Lab. The documentation provides information on the latest updates, authors, and installation process.

Bullet Points

• •Interactive deep learning book with code, math, and discussions
• •Implemented with PyTorch, NumPy/MXNet, JAX, and TensorFlow
• •Adopted at 500 universities from 70 countries
• •New JAX implementation available with new topics of reinforcement learning, Gaussian processes, and hyperparameter optimization
• •New API for implementation and new topics like generalization in classification and deep learning, ResNeXt, CNN design space, and transformers for vision and large-scale pretraining
• •Join to improve ongoing translations in Portuguese, Turkish, Vietnamese, Korean, and Japanese
• •Slides, Jupyter notebooks, assignments, and videos of the Berkeley course can be found at the syllabus page
• •Each section is an executable Jupyter notebook, where you can modify the code and tune hyperparameters to get instant feedback
• •Interactive learning experience with mathematics, figures, code, text, and discussions
• •Discuss and learn with thousands of peers in the community through the link provided in each section
• •Can be used as a textbook or a reference book
• •BibTeX entry for citing the book is provided at the end of the documentation.