Course site: https://deeplearning.cs.cmu.edu/F25/index.html.
Why I’m following it: I wanted a course that doesn’t skip the linear algebra step in the middle. The CMU lectures are tougher than the average MOOC, but the homeworks are the part I’d actually recommend - they build a small autograd from scratch before letting you touch PyTorch.
Currently on lecture 12 (CNNs / weight sharing). Going slower than I planned because I keep stopping to re-derive things on paper.