Engineer GPT Knowledge within a project
Project description
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'.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for knowledge_engineer-0.7.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9ee4989f13143ed2e078895ad142ba4e532ca03ae5b4d7ad6dcae1a27f9d4d74 |
|
MD5 | 52180ecca371ca7cc6d1f95a6bbc2edd |
|
BLAKE2b-256 | 2c255b2468c8bbc95e2e1136b8544aad250581cd506956e1b5894ed59eae1ab6 |