Skip to content

Latest commit

 

History

History
20 lines (14 loc) · 1.12 KB

README.md

File metadata and controls

20 lines (14 loc) · 1.12 KB

Demo app using Realm as a Vector Store

This is a modified version of mongodb/template-app-kotlin-todo that showcases a local semantic search and a Retrieval-Augmented Generation with a GenAI

Note: This is based on a custom version of Realm Kotlin SDK that uses a POC of KNN search.

The Demo uses 4 recent articles from https://www.mongodb.com/developer/news/ that are not yet indexed by the GPT mode gpt-3.5-turbo at the time of this demo Local_Articles.png

Then offer two search modes: Search_Modes.png

Semantic Search:

We can search local articles (The question is embedded using text-embedding-ada-002 but the KNN search is done locally) Semantic_Search.png

Retrieval-Augmented Generation Search:

RAG_search.png The GPT search brings outdated answers (Laravel 10). Whereas the RAG search uses local articles for a more up-to-date answer GPT_Search.png