After Python / Deep learning foundations
Small network.
Big first step.
Build a neural network you can actually explain. Start with a few numbers, teach them to learn, then bring it all together in PyTorch.
Who this is for
You know Python.
Start learning models.
Comfortable with variables, lists, loops, and functions? You are ready. We introduce PyTorch and explain the little bit of slope-and-gradient math as we use it. No advanced Python, GPU, or previous deep-learning course is required.
Need a Python refresher? →The learning path
From one neuron to a working model.
Follow the order. Predict an output, run the code, change one thing, and explain what happened.
Your first neuron
Turn a few Python numbers into a prediction. Understand inputs, weights, bias, and activation before importing a framework.
How a model learns
Measure a mistake, estimate a slope, and train a one-neuron model using only Python.
Meet PyTorch & tensors
Set up a real PyTorch notebook and learn to read tensor shapes without guessing.
Let PyTorch find the gradients
Connect the hand-written update to autograd, backward(), and a complete training loop.
Build your first neural network
Train a tiny two-input network on XOR and see why a hidden layer and activation matter.
Test it, save it, make it yours
Finish a small classifier with separate training, validation, and test examples, then save and reload its learned weights.
What you will leave with
A small, complete PyTorch project.
A two-input classifier, a training loop you understand, an honest test result, and a saved model you can reload. No chatbot-scale training or expensive hardware.
Begin lesson 01 →Build-first, not black-box.
This original Room310 course takes inspiration from the bottom-up, code-first progression in Andrej Karpathy's Neural Networks: Zero to Hero. It is not his course, a transcript, or an affiliated program. Our smaller finish line is a simple classifier; his course is an optional next step.