Part 7/10:
To cement the understanding of N10's functionalities, let’s walk through the practical example of a document question-answering system.
Workflow Overview
The project consists of two main flows: upserting and querying.
Upserting Flow: This will involve loading documents, splitting them into manageable chunks, transforming those chunks into vectors using an embeddings model, and storing the data in a suitable vector storage system.
Chat Flow: Users will enter their queries, which will then be converted into vectors for a semantic search against the previously stored embeddings. The language model will generate responses based on these queries.