Terminal Assistant

This is a simple, streaming chatbot that can be powered by either OpenAI or Ollama. For OpenAI, you can select from GPT-3.5-Turbo or GPT-4-Turbo models from within the app. The estimated cost is displayed at the bottom of every response. To keep costs down, restart conversations when topics change with the !thanks command. For Ollama, you just run the app with --ollama argument after setting your model name in .env. The default is "gemma:2b".

Installation and Setup

1. Python: Ensure Python 3.10 or higher is installed on your system.

2. Dependencies: Install the required Python packages with the following command:

   pip install openai tiktoken python-dotenv

   or

   pip install -r requirements.txt

3. API Key: Set your OpenAI API key, Ollama Model, and optionally the Assistant Name in .env file like this:

   OLLAMA_MODEL=gemma:2b
   OPEN_AI_KEY=sk-demo
   ASSISTANT_NAME="Terminal Assistant"

4. Identity: Modify the 'identity' variable within the code to change the bot's identity if you wish.

5. Run the Chatbot: Start the chatbot with the following command for OpenAI by default:

   python terminal-assistant.py

   and use the following command for Ollama:

   python terminal-assistant.py --ollama

Note: This bot uses the newest OpenAI python package and therefore the newer API call. If it's not working for you, you may need to update the package with pip install --upgrade openai.

Usage

• 💬 Chat: Simply type your message and press enter to chat. The conversation history is maintained until you wipe it with !thanks.

• ⌨️ Multi-line Input: Use !code to enter multi-line input mode. This is useful for providing code samples.

• 🧠 Clear History: Use !thanks to clear the chat history.

• 🔁 Model Selection: Use !gpt3 or !gpt4 to select the model.

• 👋 Quit: Use !exit to quit the application.

Screenshots

OpenAI

Ollama