AI Chatbot Project

This is for my read through of Building Your OwnApplicatoins with Local AI Models on a Mac

• Currently up through chapter 8

I will record differences here:

• Using vscode installed via official installer and also cursor

• Ollama already installed via curl rather then brew (manages its own updates)

• Using gemma4:12b-mlx instead of gemma3:4b. (But i have both pulled and plan to do comparisons).

• Also compare qwen3.6:35b-mlx and others .

  • Use uv run (or use activate, doesnt matter)
  • use uv add instead of pip, so much of chapter 6 doesnt apply

Usage

• Install uv.
• Install Ollama, make sure it is running (see below)
• uv sync (not really needed, but nice to get all that out of the way)
• use uv run for everything.

Tips

To run streamlit apps: uv run streamlit run app.py Yeah its clunky.

Set Up Ollama

Pull the models you wish to use. The app discovers them automatically via ollama.list (preferred default: gemma3:4b when installed).

For example:

ollama pull gemma3:4b

Confirm installed models:

ollama list

File structure

• ch10 : files created for chapter 10. not sure why we put this in a sub folder
• experiments ; various small scripts to experiment with ollama , streamlit etc.

Main app

The main app is 'chatbot.py'

uv run streamlit run chatbot.py