Autograd, Loss, Backpropagation, and Training Loops

demystify how neural networks learn.

Units

1. Unit 03.00: Forward pass and loss
2. Unit 03.01: Gradients and `requires_grad`
3. Unit 03.02: Backpropagation intuition without heavy derivation
4. Unit 03.03: Optimizer step and zeroing gradients
5. Unit 03.04: Building a minimal training loop
6. Unit 03.05: Saving metrics and checkpoints
7. Unit 03.06: Project step: train a tiny model end to end

Module work

• Activity
• Knowledge check
• Summary