d2l zh

The book covers the complete arc of modern deep learningfrom linear neural networks and multilayer perceptrons through convolutional networks, recurrent architectures, attention and transformers, to computer vision and NLP applicationswith interactive, runnable code examples throughout.

The textbook's distinctive approach integrates mathematical derivations, conceptual explanations, and executable code in a unified format where theory and implementation reinforce each other.

Every concept is illustrated with code that readers can run, modify, and experiment within PyTorch, TensorFlow, or MXNetproducing immediate feedback that builds intuition alongside mathematical understanding.

The Chinese adaptation preserves this interactive format while providing culturally familiar examples and Chinese-language mathematical notation conventions where they differ from English-language standard.

Chinese university students in computer science and engineering programs, self-directed learners entering deep learning from adjacent fields, and professionals at Chinese technology companies building deep learning capability use d2l-zh as a primary or supplementary textbook.

The open-access format (freely available online) and the quality of the original d2l textdeveloped by Mu Li, Alex Smola, and colleagues at Amazon and Carnegie Mellonhave made this one of the most widely used deep learning textbooks globally, with the Chinese version extending that reach into the world's largest technology talent market.