Barry-Bee - Eco Virtual-Assistant (Prototype)

A proof of concept prototype of a virtual assistant using a custom fine-tuned version of the H4rmoniousCaramel the ecolinguistic large language model from https://theh4rmonyproject.org/

For testing purposes, the LLM is hosted locally on my machine using a python flask server and connection is then established from the Unity game engine. Also due to using outdated hardware, I could only train the model  with just less than 1/8th of it's dataset, so responses are not as in-depth/technical as the fully trained model can provide.  It's also worth considering that for demostational purposes, a fully trained model would struggle to run efficiently on a flask sever.

Regardless, Barry Bee can receive prompts, process and provide responses based on the limited data the model has been trained with to a reasonable/acceptable state all at runtime inside the games engine. Showcasing the possible implementations of the language model in 3rd party applications such as home assistants, video games, interactive art, TTS services, etc.

Github Available Here: https://github.com/dango17/EcoBeeLLM