This repository is for profiling, extracting, visualizing and reusing generative AI weights to hopefully build more accurate AI models and audit/scan weights at rest to identify knowledge domains for risk(s).
Implementation of Qformer from BLIP2 in Zeta Lego blocks.
