Skill
PyTorch basics
Tensors, autograd, a training loop you could rewrite from memory.
Do these first
What you will be able to do
- Create tensors and run a forward pass
- Write a training loop with loss.backward() and optimizer.step()
- Move a model to GPU when available
Curated resources
Start with the pick, then branch out
Scores are LearnPath editorial opinions (clarity, cost, freshness, project density) — not an objective ranking. Every YouTube item is embedded officially after a click-to-load facade.
Official docs
LearnPath pickdocumentationintermediate
PyTorch tutorials
PyTorch3h
Official basics track. Run locally or in Colab.
Score 95
Open
25h 37mLearnPath pickMore curatedyoutube coursebeginner
PyTorch for Deep Learning & Machine Learning – Full Course
freeCodeCamp.org25h 37m
Twenty-five hours of PyTorch with a project-driven structure — tensors, autograd, training loops, datasets, transfer learning, then three real projects including computer vision. Effectively a full deep-learning curriculum in one video.
Score 87
Watch
25h 37mMore curatedyoutube coursebeginner
Learn PyTorch for deep learning in a day. Literally.
Daniel Bourke25h 37m
The 'Learn PyTorch for deep learning in a day' course — nearly 26 hours, with notebooks, exercises and extra-curricular problems. A structured alternative to the freeCodeCamp course with a strong emphasis on doing rather than watching.
Score 85
Watch
4h 36mMore curatedyoutube coursebeginner
Deep Learning With PyTorch - Full Course
Patrick Loeber4h 36m
A four-and-a-half-hour PyTorch course with a clean linear progression — tensors, autograd, training, CNNs, RNNs — useful when the 25-hour course feels like too much to schedule.
Score 80
Watch