This repo is a fork of the AI and I Hackathon: Co-Creativity between Humans and LLMs, organized by fortiss and the partners: Start2 Group, LMU Munich, DFKI, and Humane AI Net.
Ensure you have Docker and Docker Compose installed on your machine. You can download them from:
To run locally you will need docker installed!
You also need the following environment variables set:
OPENAI_API_KEY
- an OpenAI access key.SSL_KEY
- a valid OpenSSL certificate keySSL_CERTIFICATE
- a valid OpenSSL certificate
Create a .env file where you store your certificates and credentials in the root folder.
-
Create a directory for the SSL certificate:
mkdir certs
-
Generate a self-signed SSL certificate and key:
openssl req -x509 -nodes -days 365 -newkey rsa:2048 -keyout certs/selfsigned.key -out certs/selfsigned.crt
In the .env file, you can reference the respective files.
The prompts for the agents are located inside task-examples.
The frontend for the hackathon is based on our poetry task, rewritten in ReactJS. The main idea of the task is that the user submits the theme/objective at the of the mindmap and then collaborotively (with AI) extends the mindmap. During the process. From the mindmap, our agent will create a text.
At the moment, the system supports two models gpt4-turbo and gpt4-o, which is located inside the folder model_template/models.
You can change between the two models by changing the value of the variable ai_model
between OpenAIImageModel()
and OpenAIModel()
in the file model_template/model.py.
If you have time and want to add your own models to the system, feel free to do so by following the template located in the file basemodel.py and use the already existing model files as guidance.
The task_template
and model_template
folders contain template applications for deployment on the AI Builder infrastructure.
They indicate how to implement a model and a task and supply most infrastructure necessary to minimise the requirements of a user to adopt their code.
Details are provided in the respective README files.
To build the application, run:
docker compose -f docker-compose_poetry.yaml build
from the root directory.
To run the application, run:
docker compose -f docker-compose_poetry.yaml up --force-recreate
from the root directory.
To stop the application, run:
docker compose -f docker-compose_poetry.yaml down
from the root directory.