The Deep Learning Dihedral

Move from classical ML into the world of neural networks. Learn the mechanics of tensors, gradients, and backpropagation before building your first models.

What You'll Learn

• Tensors and matrix operations as the foundation of deep learning
• PyTorch fundamentals: autograd, training loops, and model building
• Your first feedforward neural network from scratch
• Graph neural networks and proper evaluation metrics
• When deep learning helps (and when it doesn't)

Why This Wall Matters

Dihedrals demand precision. You wedge yourself in and make small, controlled moves. Deep learning is the same: before you can build transformers or diffusion models, you need to master the fundamentals. This wall builds that foundation — tensors, gradients, and simple networks — so the advanced routes feel natural.

Prerequisites

Complete W06 (The Machine Learning Offwidth) first. You should be comfortable with sklearn workflows before moving to PyTorch.