LLM Primer

Exercises

These exercises check whether the PyTorch names still point to visible computation.

Answer it first, then check.

Answer it first, then check.

Answer it first, then check.

Answer it first, then check.

Answer it first, then check.

Answer it first, then check.

Answer it first, then check.

Written practice

  1. Rewrite a five-line PyTorch training step in plain English.
  2. Explain why zero_grad() normally appears before backward().
  3. Describe the difference between a module and a parameter.
  4. Explain why the course introduced PyTorch after losses, gradients, and optimizers.