This project provides a versatile chat interface for interacting with Large Language Models (LLMs), targeting users who want offline, multi-modal LLM communication on various hardware, including resource-constrained devices like Raspberry Pi. It offers command-line, browser-based, and voice-controlled interaction modes.

How It Works

The system leverages llama.cpp as the LLM backend, enabling local inference without internet connectivity. It supports multiple front-end interfaces: a standard terminal console, a browser-based console using JavaScript, and a voice interface that integrates speech recognition and text-to-speech. The architecture is designed for flexibility, allowing these interfaces to run on both standard computers and embedded systems like the Raspberry Pi.

Quick Start & Requirements

• Install: Clone llama.cpp, build it (make), download a compatible GGUF model (e.g., openchat-3.5-0106.Q2_K.gguf), and run the llama.cpp server.
• Prerequisites: Linux/Mac/Windows terminal, or a browser. A Raspberry Pi 5 is recommended for offline operation.
• Setup: Requires cloning and building llama.cpp and downloading an LLM model.
• Docs: llama.cpp

Highlighted Details

• Offline LLM operation on resource-constrained devices.
• Multiple interaction modes: terminal, browser, voice.
• Designed for Raspberry Pi compatibility.

Maintenance & Community

The project relies on llama.cpp, which is actively maintained by Georgi Gerganov and a community. Specific community channels for susi_chat are not detailed in the README.

Licensing & Compatibility

The project's licensing is not explicitly stated in the README. llama.cpp is typically distributed under the MIT License, which is permissive for commercial use.

Limitations & Caveats

The README focuses primarily on setting up the llama.cpp backend and does not detail the installation or functionality of the susi_chat interfaces themselves. Model compatibility and performance will vary significantly based on the chosen LLM and hardware.