Our work on developing AI sensitivity to the changing mental states implied in the word flows in texts and trasnscription resulted in a Retrieval-Augmented Generation (RAG) adjunct to a local Mistral LLM server.

Recently we have been exploring the use of decription vectors in carefully designed AI prompts which give a set of parameters that characterise mental states and transitions – differential mental states. We used mistral LLM and the designed prompt to process some 500 hours of transcripts of Buddhist study and practice sessions, together with analysis of selected core mahamudra and other Buddhist text

The form below gives access access to our ‘proof of principle’ RAG-AI server.

The server is very mahamudra centric. If you ask it a question you will soon see what I mean.