Editor's top PyTorch books
3Deep Learning with PyTorch
Eli Stevens, Luca Antiga, Thomas Viehmann · Manning · 542 pages · ~14 hrs to read
The most comprehensive PyTorch book — covers tensor operations, autograd, CNNs, RNNs, and a full end-to-end cancer detection project. Written by a PyTorch core contributor. Free chapters available online.
Machine Learning with PyTorch and Scikit-Learn
Sebastian Raschka, Yuxi Liu, Vahid Mirjalili · Packt · 775 pages · ~19 hrs to read
Sebastian Raschka's update to his classic ML book — covers classical ML with scikit-learn through deep learning with PyTorch and transformers in one volume. Best for engineers who want the full ML picture.
Deep Learning for Coders with fastai and PyTorch
Jeremy Howard, Sylvain Gugger · O'Reilly · 676 pages · ~17 hrs to read
The book version of the fast.ai course — trains working vision, NLP, tabular, and collaborative-filtering models before explaining the underlying PyTorch mechanics. The strongest pick for programmers who want results before theory.
PyTorch Books for Beginners
3Deep Learning with PyTorch
Eli Stevens, Luca Antiga, Thomas Viehmann · Manning · 542 pages · ~14 hrs to read
The most comprehensive PyTorch book — covers tensor operations, autograd, CNNs, RNNs, and a full end-to-end cancer detection project. Written by a PyTorch core contributor. Free chapters available online.
Deep Learning with PyTorch Step-by-Step
Daniel Voigt Godoy · self-published · 380 pages · ~10 hrs to read
A three-volume series (Fundamentals, Computer Vision, Sequences & NLP) that builds autograd, nn.Module, and training loops up line by line in a deliberately conversational style with minimal notation. The gentlest on-ramp for engineers intimidated by academic PyTorch books.
Deep Learning for Coders with fastai and PyTorch
Jeremy Howard, Sylvain Gugger · O'Reilly · 676 pages · ~17 hrs to read
The book version of the fast.ai course — trains working vision, NLP, tabular, and collaborative-filtering models before explaining the underlying PyTorch mechanics. The strongest pick for programmers who want results before theory.
Intermediate & Advanced PyTorch Books
7Machine Learning with PyTorch and Scikit-Learn
Sebastian Raschka, Yuxi Liu, Vahid Mirjalili · Packt · 775 pages · ~19 hrs to read
Sebastian Raschka's update to his classic ML book — covers classical ML with scikit-learn through deep learning with PyTorch and transformers in one volume. Best for engineers who want the full ML picture.
Programming PyTorch for Deep Learning
Ian Pointer · O'Reilly · 216 pages · ~5 hrs to read
Focused on the practical engineering side of PyTorch — transfer learning, deployment with Flask and TorchServe, and cloud deployment on AWS/GCP/Azure. Strong on the production path.
Build a Large Language Model (From Scratch)
Sebastian Raschka · Manning · 366 pages · ~9 hrs to read
Implements a GPT architecture entirely in PyTorch from scratch — the most direct path to understanding how modern LLMs work at the tensor level. Essential reading for anyone fine-tuning or deploying language models.
PyTorch Pocket Reference
Joe Papa · O'Reilly · 265 pages · ~7 hrs to read
A syntax and pattern reference rather than a teaching text — data loading, custom training loops, GPU/TPU acceleration, and deployment to AWS, GCP, or mobile in one compact volume. Keep it next to the keyboard, not on the reading list.
Natural Language Processing with PyTorch
Delip Rao, Brian McMahan · O'Reilly · 258 pages · ~6 hrs to read
Pre-dates the transformer era but still the clearest treatment of building NLP models — embeddings, RNNs, CNNs for text — directly in PyTorch tensors rather than through a high-level library. Useful for understanding what HuggingFace abstracts away.
Inside Deep Learning
Edward Raff · Manning · 598 pages · ~15 hrs to read
Explains why each PyTorch component exists, pairing dozens of runnable PyTorch examples with plain-language math and no proofs. A good bridge for engineers who've followed tutorials but can't yet explain what a layer is actually computing.
Modern Computer Vision with PyTorch, 2nd Edition
V Kishore Ayyadevara, Yeshwanth Reddy · Packt · 747 pages · ~19 hrs to read
Over 40 vision applications in PyTorch, updated through CLIP and Stable Diffusion — object detection, segmentation, GANs, multimodal models. The most current computer-vision-specific PyTorch book on the list.
PyTorch resources
Learn next