Knowledge based prompting for GPT-4

Generated Knowledge Prompting | Prompt Engineering Guide<!-- --> I came over this website, the approach looks similar to few shot prompting if I’m not wrong? Also I’m working on a text-to-video use case by exploring open source models… so I have used a RAG system to extract the process from a document and this will be passed to the gpt-4 model which will inturn generate prompts to be fed to text-to-video models. The issue is some of the words it uses in the prompts , specifically name of the products, the t2v models are not able to understand so I wanted the gpt-4 model to describe the shape ,size features of the product instead of just naming it. The features are not available in the pdf, it has only product name so I was hoping to use knowledge based prompting where it will use its own database to answer it. Can someone help me out , im not sure how to implement it. Found this github repo but some files seem to be missing- GitHub - liujch1998/GKP