We improve generated answer question correcctness from 61% to 70% by setting up a system to dynamically generate prompts with example cases most relevant to the input question.
Built with Pongo's semantic filter for search relevancy and griptape.ai's frameworks for quick, modular code
See the blog post for more: