PIP-BOY LLM

Vault-Tec Local AI Terminal - Run local LLMs with a retro Fallout PIP-BOY themed interface.

Installation

pip install pip-boy-llm

Optional Dependencies

# For 4-bit quantization (Mistral 7B)
pip install pip-boy-llm[quantization]

# For PDF file support
pip install pip-boy-llm[pdf]

# For Windows readline support
pip install pip-boy-llm[readline]

# Install all optional dependencies
pip install pip-boy-llm[all]

Quick Start

1. Run the setup wizard (recommended for first-time users):

   pip-boy-setup
   

2. Start the terminal:

   pip-boy-llm
   

3. Select a model and start chatting!

Models

HuggingFace Login

Some models require HuggingFace authentication:

1. Create an account at huggingface.co
2. Generate a token at huggingface.co/settings/tokens
3. Run pip-boy-setup and enter your token

License Agreements

Llama and Gemma models require accepting license agreements:

• Llama 3.2: Accept license
• Gemma 3: Accept license

Commands

File References

Include file contents in your messages using @:

> Explain this code: @main.py
> Summarize these files: @src/app.py @src/utils.py
> Review @"path with spaces/file.py"

Commands Reference

pip-boy-llm

Main terminal interface. Select a model and start chatting.

pip-boy-llm

pip-boy-setup

Setup wizard for first-time configuration:

• Checks dependencies (PyTorch, Transformers, etc.)
• Configures HuggingFace authentication
• Verifies model access
• Creates config directory

pip-boy-setup

pip-boy-update

Check for package updates:

pip-boy-update

Configuration

Config files are stored in ~/.airllm/:

• config.yaml - User preferences
• history.yaml - Chat history

System Requirements

• Python 3.9+
• CUDA-capable GPU recommended (CPU mode available)
• 4GB+ VRAM for 1B models
• 8GB+ VRAM for Mistral 7B (4-bit)

License

MIT License