Tiny Notepad with Ollama

A sleek local notepad app powered by Ollama. Write notes, chat with open-source language models, and tweak generation settings. Built with a minimal Tkinter-based GUI. The goal of the project is to be an accessible, light weight version of open-webui capabilities available via a notepad app.

What is Tiny Notepad?

Tiny Notepad is a Python desktop app that allows you to:

• Write and save timestamped notes
• Load previous notes from a sidebar
• Interact with any locally installed Ollama model
• Adjust LLM generation parameters like temperature, top_p, top_k, etc.

Ideal for journaling, note-taking, or exploring how LLMs respond to prompts — all running entirely on your machine.

---

Features

• Clean Tkinter GUI

• Local-first: works offline with Ollama

• Sidebar to browse saved notes

• Real-time interaction with models (via Ollama API)

• Editable generation settings:

  • temperature
  • top_p
  • top_k
  • repeat_penalty
  • presence_penalty
  • frequency_penalty
  • stop sequences

---

Installation

1. Install Python (>=3.7)

2. Install Tiny Notepad

pip install tiny-notepad

3. Install ollama (required) Download Ollama and follow platform-specific installation instructions.

4. Run a model via Ollama

Before using the app, make sure a model is pulled and running:

ollama pull llama3
ollama run llama3

OR

ollama serve

---

Usage

Once installed via pip, launch the app using:

tiny-notepad

The app will auto-check for Ollama and attempt to start ollama serve if it's not already running.

---

Saved Notes

All notes are saved in a notes/ directory, auto-named like:

notes_2025-05-18_14-33-21.txt

You can browse and reopen them from the built-in sidebar.

---

Generation Parameters

Parameter	Description	Default
temperature	Controls randomness (0 = deterministic, 2 = wild)	0.7
top_p	Nucleus sampling (probability threshold)	0.9
top_k	Limits to top-k tokens	40
repeat_penalty	Penalizes repeated tokens	1.0
presence_penalty	Encourages new topic exploration	0.0
frequency_penalty	Penalizes frequent tokens	0.0
stop	Comma-separated stop sequences to end generation	(custom)

All parameters can be adjusted in the GUI in real time.

---

Tip

Pull and try other models using Ollama:

ollama pull mistral
ollama pull gemma

They will appear in the model dropdown after restarting the app.

---

License

MIT License © 2025 Ronald Wilson