datawhalechina/leedl-tutorial

Comprehensive coverage spans foundational optimization techniques (local minima, saddle points, adaptive learning rates) through advanced topics including CNNs, Transformers, GANs, diffusion models, reinforcement learning, and interpretable AI, with detailed mathematical derivations to lower entry barriers. The tutorial supplements lecture content with additional deep learning knowledge and includes accompanying homework code implementations across all major chapters. Designed as a structured pathway for practitioners progressing from core theory to specialized domains like adversarial robustness, transfer learning, and continual learning.

16,394 stars.