udacity/deep-learning-v2-pytorch — reverse-engineered prompt

Build me a beginner friendly deep learning learning repo using Jupyter notebooks and PyTorch. I want it to feel like a Udacity style nanodegree workspace, with clear folders for lessons and projects, starter code, explanations, and runnable exercises.

Cover the main topics people expect, intro neural networks, gradient descent, PyTorch basics, convolutional networks, transfer learning, autoencoders, style transfer, recurrent networks, Word2Vec, sentiment analysis, attention, GANs, DCGANs, CycleGANs, batch normalization, and weight initialization. Include project notebooks for bike rental prediction, dog breed classification, TV script generation, and face generation.

Please make the notebooks educational, not just finished code. Add markdown instructions, TODO sections, helper functions, and expected outcomes so a learner can work through them. Include a simple README that explains what each section is for and how to set up the Python environment with conda and the requirements file. Look up current docs online if you need to.