Catalog RAG

The Catalog RAG provides a quick search and retrieval of the catalog used for Tokkio retail reference workflow. It provides APIs to manage Catalog data and and perform various operations such as searching for items by name or ID, filtering items based on certain criteria, and retrieving distinct values for a given field.

The Catalog RAG creates a vectorstore from the uploaded json-based catalog using a specified embedding model. It utilizes the FAISS library for vector storage and HuggingFace models for generating embeddings. Queried data is searched and retrieved from the vectorstore.

Information regarding understanding and modifying the catalog can be found in Catalog Customization section.

API Guide

See the Catalog RAG API Guide for more information.