Interactive Notebook 4: Deep Learning, Artificial Neurons & Language Models

Simulating a Single Artificial Neuron (Perceptron)

Biological brains inspired the artificial perceptron node. In Experiment 1, you write a Python class `ArtificialNeuron` equipped with weights (`w1`, `w2`) and a bias (`b`).

When fed input readings for soil dryness and crop urgency, the neuron computes the weighted sum and fires a binary activation decision on whether to switch on the village irrigation water pump.

Miniature Language Model: Predicting Next Words like ChatGPT

How do Large Language Models (`LLMs`) like ChatGPT, Gemini, or Claude generate human-like sentences? At their core, they analyze vast text corpora and predict the most statistically likely next word for any sequence.

In Experiment 2, your computer reads a short village story, counts exact word follow-patterns (`follow_counts`), and automatically generates fluent new sentences from scratch. You can even type your own sentences in any language to watch the AI learn your writing style!

How to Open and Run Notebook 4

There are two easy ways to run the notebook: on your own computer if Jupyter is installed, or free in the browser with Google Colab, where nothing needs to be installed. The steps for Google Colab:

1. Download the notebook file with the button below.
2. Open colab.research.google.com in your browser and sign in with a Google account.
3. Click File, then Upload notebook, and choose the downloaded file loktech-notebook-1.ipynb.
4. Read each text cell, and run each code cell by pressing the play button on its left (or Shift+Enter).
5. Run the cells from top to bottom. Change the examples and run them again; nothing can break.