Knowledge Engineer

Command Line Program allowing for the Engineering of Knowledge for LLM's.

OpenAI, Mistral, Anthropic are supported, I am working on OLLama for local LLM execution. The program can be extended by modifying the "Knowledge Engineer Prompt File" Domain Specific Language (DSL).

(That is a fancy way to say: "by modifying the syntax of the .kepf files.")

Quick Start

Preparation linux / mac

    mkdir ke
    cd ke
    python3 -m venv .venv
    source .venv/bin/activate

Preparation Dos/Windows

    mkdir ke
    cd ke
    python3 -m venv .venv
    .venv\Scripts\activate

Installation

    pip install knowledge-engineer

Create New Project

    knowledge_engineer --create=snake

Setup Project

    cd snake
    edit ke_process_config.env

Enter Your API Keys in the ke_process_config.env file:

OPENAI_API_KEY='<Your Open API Key>'

execute example Project

    knowledge_engineer --execute

Run Your Generated Snake Game

    python Code/snake_game.py

This proves knowledge-engineer is running. But the game will probably not work, as Chat GPT 3.5 Turbo only gets it right about 1/4 of the time.

Please see getting Started below for an explanation of how to use knowledge_engineer.

Further Documentation

For Users

• The Getting Started document contains a good place to start. It is a PDF file that is a good introduction to the example project you just created and ran. It is also available in HTML format.

• The KEPF Language Reference document is a good place to look into the options and syntax of the Domain Specific Language (DSL) used to create the prompts for the LLMs. It is also available in HTML format.

For Developers

• Aditionaly, the Release A Version explains how to update PiPy (pip install) directory with future versions of the knowledge-engineer package.

For Philosophers

• The Why Knowledge Engineer document explains why I feel that LLMs need 'Knowledge Engineer'.