Learning Path 01

PyTorch: Zero to Hero

Learn the machinery behind deep learning by building upward from tensors and gradients to attention, training and a small language model.

A practical bottom-up path for understanding what PyTorch is doing rather than memorizing framework calls.

Foundations 11 chapters Sequential or reference
Start at chapter 00 Use with AI
Complete curriculum

Chapters

Work through the path in order, or jump directly to the mechanism you need.

  1. 00Build a Neural Network From Scratch in PyTorch Without nn.Module
  2. 01PyTorch Attention Shapes: Q, K, V, Multi-Head Attention Masks and Transformer Dimension Errors
  3. 02PyTorch Autograd Debugging: requires_grad, detach, backward() and NaN Gradients
  4. 03PyTorch CNN Shape Errors: Conv2d Output Sizes, Channels, Flatten Bugs and How to Debug Them
  5. 04PyTorch DataLoader Performance: num_workers, pin_memory, Prefetching and Why Your GPU Is Waiting
  6. 05PyTorch Model Not Learning? A Systematic Debugging Guide
  7. 06PyTorch nn.Module Explained: Missing Parameters, state_dict, Buffers and Registration Bugs
  8. 07PyTorch Performance Debugging: CUDA OOM, Slow Training, GPU Utilization and torch.compile
  9. 08PyTorch Tensor Shapes: Broadcasting, Reshape, View, Permute and the Errors That Waste Your Time
  10. 09What Are We Actually Doing?
  11. 10Build a Small GPT-Style Language Model From Scratch