In this video, we're going to learn how to do Search-Based RAG on a podcast trasnscript. Instead of doing vector search to retrieve appropriate text chunks, we'll use Full-Text Search with DuckDB. We'll also use GLiNER to extract key words from the user's prompt.
#gliner #duckdb #llamacpp
Import code - [ Ссылка ]
Search code - [ Ссылка ]
Simon Willison's Blog - [ Ссылка ]
GLiNER - [ Ссылка ]
GLiNER model - [ Ссылка ]
llama.cpp - [ Ссылка ]
![](https://s2.save4k.ru/pic/mLYHsrSxgpU/maxresdefault.jpg)