This is a basic sequential system that is used to extract knowledge from pdfs by exploiting locally run models. Although the repo is highly tailored, it was made to run on older GPUs (with small context windows) for text extraction, generation and (mermaid) schema creation.