KARL

Künstlicher Assistent repräsentiert durch LLMs

---

Karl is a small robot living inside a tube television. With the help of a library called Ollama, he can access artificial intelligences. Primarily, Llava and Llama2 are used, which operate in both image and text-based contexts. Users can have documents processed, get image descriptions, and analyze current websites through Karl. Llama2 has a large database that can be accessed through Karl.

---

🖥️ Features

• 🤖 - Local artificial intelligence
• 💬 - Chatbot function
• 🗣️ - Speech-to-text function
• 🖼️ - Image scanning
• 📄 - Document scanning
• 📷 - Webcam scanning

KARL's intelligence operates locally and can function without internet connectivity in theory; it can also be switched to other AI models and process various types of documents.

[Examples]

KARL is utilizing information from a website as a source. Check out the difference between the responses.

KARL is utilizing information from an image as a source. Based on that he can provide answers.

🕓 Planned Upcomming features

• 🔗 Expanded File Type Integration: Incorporate additional document formats such as docx or excel.
• 🤖 AI-Powered Coding: Utilize KARL as your primary tool for solving daily coding challenges.

📝Requirements

• Ubuntu version 22.04: 🐧🖥️2️⃣2️⃣.0️⃣4️⃣
• Minimum 20GB RAM: 💾2️⃣0️⃣GB
• Nvidia GPU for Ollama: 🎮💻🔥
• Minimum 10-core CPU: 💻🔟
• Google Chrome as the browser: 🌐🔍

📜 ⬇️ Installation guide

This guide will walk you through the installation process step by step.

[Installation guide]

Step 1: Update Package Manager

sudo apt update

Step 2: Install Required Packages

sudo apt install curl git nodejs npm python3 python3-pip wget

Step 3: Install Additional Packages

sudo apt full-upgrade
sudo apt install npm
sudo apt install google-chrome-stable
sudo apt-get install -f

Step 4: Install Python Libraries

pip install flask flask-cors langchain

Step 5: Install Ollama

sudo curl https://ollama.ai/install.sh | sh

Step 6: Pull required LLMs

ollama serve
ollama pull llava
ollama pull llama2

Step 7: Clone KARL Repository

cd Desktop
git clone https://github.com/vw-wob-it-edu/KARL.git
cd KARL

Step 8: Install Python Requirements

pip install -r requirements.txt

Step 9: Reboot System

sudo reboot

Step 10: Start server

cd KARL-Server #where project is located
chmod +x server-start.sh
./server-start.sh

👥 The team

• Marvin-VW
• Daniel-VW
• niklasprg

💼 License

MIT License

Copyright (c) 2024 KARL

---

Services

Fonts:

• Roboto Mono: Used for code blocks and monospace text.
• Grape Nuts: Presumably used for specific stylized text elements.

External Libraries and Scripts:

• Core CSS and JS from pyscript.net: custom styling and functionality.
• Font Awesome: Icon library for scalable vector icons.
• Webcam Easy: Script for handling webcam functionalities.

Dependencies

Node.js/Express:

• Express: Framework for building web applications with Node.js.
• Body Parser: Middleware for parsing incoming request bodies.
• Axios: Promise-based HTTP client for making requests.

Python/Flask:

• Flask: Micro web framework for Python.
• Flask CORS: Extension for handling Cross-Origin Resource Sharing (CORS).
• Langchain: Custom library for language processing tasks.

Scripts

Web Page Loading:

• langchain.document_loaders.WebBaseLoader for loading web pages.
• langchain.text_splitter.RecursiveCharacterTextSplitter for text splitting.

Embedding and Storing:

• langchain.vectorstores.Chroma for vector storage.
• langchain.llms.Ollama is used for language modeling.

Flask App:

• Handles API endpoints for text processing and file handling.

© Copyright KARL 2024