A simple script in python to let two GPT4all models talk together about Universe and Life. One of the two models is simply instructed to generate a follow up question for the answer of the other model, after a human introduce some initial topic.
Download GPT4All here: https://www.nomic.ai/gpt4all and install it on two Linux hosts.
The GPT4all models runs on hosts with "Enable API" set to true, and SSH tunnels make sure they are reachable. Conversation sentences are stored in a PostgreSQL database with pgvector, and the sentences are embedded with "all-MiniLM-L6-v2" on a 384-dimensional space for later retrieval.
For the sample in conversation table, a human introduced the topic with "Measure the Universe", then the debate went on with "Reasoner v1" questioning on a RTX 3050 and "bartowski/gemma-2-9B-it-GGUF" answering on a RTX 3090.
The code was created with Vibe Programming with help of ChatGPT 4.5.