DL-Hub

Why we included this project

DL-Hub teaches deep learning through runnable PyTorch code rather than passive slides. Its 339 lessons cover the field broadly, from computer vision and NLP to graph networks, point clouds, generative models, LLMs, and multimodal systems, and each one follows the same shape: define the problem, bring the data, build the model, train it, verify it. A shared Model Zoo holds hundreds of local PyTorch implementations behind one interface, so swapping architectures and comparing behavior takes a single line. Everything runs offline on synthetic data with a common CLI, which suits self-learners who want to actually modify and train models, as well as instructors assembling reproducible course material. Beginners get staged learning paths and weekend sprints, while advanced users can audit the 80 curated zoo sources or work through the 50+ LLM paper notes.