What You Need

Required

  • Python basics — variables, functions, loops, classes
  • Python 3.9+ — installed on your machine
  • Git — to clone the repository
  • 2-4 hours — split across multiple sessions is fine
No GPU required. This model is tiny (~100K parameters) and trains in minutes on any laptop CPU.

Optional

  • Hugging Face account — free, to deploy your finished model online

Sign up at huggingface.co when you're ready to deploy.

Helpful but NOT Required

  • PyTorch experience — we'll explain everything
  • Linear algebra — we'll cover what you need
  • Calculus — backpropagation is explained intuitively
  • ML experience — this tutorial assumes none

Quick Self-Check

Can you read this Python code?

python
1class Calculator:
2 def __init__(self, vocab_size):
3 self.vocab_size = vocab_size
4
5 def forward(self, tokens):
6 for token in tokens:
7 result = self.process(token)
8 return result

If you can roughly follow what this does (a class with methods that loops through tokens), you're ready. You don't need to know PyTorch or ML. We'll teach you.

Helpful?