NoLlama

NoLlama is a terminal-based interface for interacting with large language models (LLMs) that you can't run locally on your laptop. Inspired by Ollama, NoLlama provides a streamlined experience for chatting with models like GPT-4o, GPT-4o-mini, Claude 3 haiku, Mixtral, LLaMA 70B, and more, directly from your terminal.

While Ollama offers a neat interface for running local LLMs, their performance and capabilities often fall short of these massive models. NoLlama bridges this gap by allowing you to interact with these powerful models using a lightweight terminal UI, complete with colorful markdown rendering, multiple model choices (currently 15+ models), and efficient memory usage.

Features

• Multiple Model Choices: Switch between various LLMs like GPT-4o, GPT-4o-mini, Mixtral, LLaMA 70B, Claude 3 haiku and more.
• Neat Terminal UI: Enjoy a clean and intuitive interface for your interactions.
• Colorful Markdown Rendering: Unlike Ollama, NoLlama supports rich text formatting in markdown.
• Low Memory Usage: Efficient memory management makes it lightweight compared to using a browser for similar tasks.
• Easy Model Switching: Simply type model in the chat to switch between models.
• Clear Chat History: Type clear to clear the chat history.
• Exit Prompt: Type q, quit, or exit to leave the chat.
• Default Mode: NoLlama runs in standard mode by default—just type nollama in the terminal to start.
• Experimental Feature: Enable live streaming of output with the --stream flag (unstable).
• Anonymous and Private Usage: Use torsocks to route all traffic through the Tor network for privacy.

Installation

1. Download the Linux binary from the releases page.

2. Make the binary executable:

   chmod +x nollama

3. Move the binary to /usr/bin/ for global access:

   sudo mv nollama /usr/bin/

4. Run NoLlama:

   Start NoLlama from the terminal by simply typing:

   nollama

   This will start NoLlama in the default mode.

Note: The provided Linux binary is compiled using Python 3.12.4 and requires glibc 2.38+ to run. This may not work on some stable distributions that ship with older packages, such as Debian. In such cases, it is recommended to build from source to ensure compatibility.

Building from Source

If you'd like to build NoLlama from source, follow these steps:

1. Clone the Repository:

   git clone https://github.com/spignelon/nollama.git
   cd nollama

2. Install Dependencies:

   You can install the required dependencies using pip: Creating a python virtual environment:

   virtualenv .venv
   source .venv/bin/activate

   pip install -r requirements.txt

3. Compile the Script (Optional):

   If you want to compile the script into a standalone executable, you can use PyInstaller:

   First set version_check: bool = False in .venv/lib/python3.12/site-packages/g4f/debug.py

   Then:

   pyinstaller --onefile --name=nollama --collect-all readchar nollama.py

4. Move the Executable to /usr/bin/:

   After compilation, move the binary to /usr/bin/:

   sudo mv dist/nollama /usr/bin/nollama

5. Run NoLlama:

   Start NoLlama by typing:

   nollama

Usage

• Switch Models: Type model in the chat to choose a different LLM.

• Clear Chat: Type clear to clear the chat history.

• Exit: Type q, quit, or exit to leave the chat.

• Default Mode: Run NoLlama without any flags for standard operation:

  nollama

Anonymous and Private Usage

For enhanced privacy and anonymity, you can use torsocks to route NoLlama's traffic through the Tor network. This ensures that all requests are anonymized and cannot be traced back to you.

Step 1: Install Tor

Debian/Ubuntu:

sudo apt update
sudo apt install tor

Arch Linux:

sudo pacman -S tor

Fedora:

sudo dnf install tor

Step 2: Enable and Start Tor

After installation, you need to enable and start the Tor service:

sudo systemctl enable tor
sudo systemctl start tor

Step 3: Run NoLlama with Tor

Once Tor is running, you can use torsocks to run NoLlama anonymously:

torsocks nollama

This will ensure that all your interactions with NoLlama are routed through the Tor network, providing a layer of privacy and anonymity.

Experimental Feature

• Streaming Mode:

  NoLlama includes an experimental streaming mode that allows you to see responses as they are generated. This mode is currently unstable and may cause issues. To enable streaming, use the --stream flag:

  nollama --stream

Todos

• Add context window
• Web interface

Contribution

Contributions are welcome! If you have suggestions for new features or improvements, feel free to open an issue or submit a pull request.

Acknowledgments

• g4f: Used for connecting to various LLMs.
• Python Rich: Used for colorful markdown rendering and improved terminal UI.

Disclaimer

NoLlama is not affiliated with Ollama. It is an independent project inspired by the concept of providing a neat terminal interface for interacting with language models, particularly those that are too large to run locally on typical consumer hardware or not available for self hosting.

License

This project is licensed under the GPL-3.0 License.