Loading...
Skip to main content

Part 2 - Document Question + Answering (DocsQA)

What we cover in part 2

In Part 1 we got introduced to Sidekicks, set-up our dev machine, and deployed the template Sidekick to Fixie.

In Part 2, you will:

  • Learn about Document Question + Answering (DocsQA).
  • Perform some queries against an existing document collection.
  • Add a document collection to our Sidekick.
  • Learn how to create a new docs collection.

DocsQA Overview​

One of the common use cases for a Fixie Sidekick is to answer questions about a collection of documents, such as web pages or PDF files. By providing a Large Language Model with information drawn from a set of documents, the Sidekick can do an excellent job answering questions and having a conversation with the user about the contents of those documents. This approach is called Retrieval-Augmented Generation, or RAG, and is a popular way of using LLMs.

Fixie Sidekicks make it easy to do this, as the Fixie service can handle all of the details of crawling documents, generating embeddings, managing a vector database, and all of the other machinery involved in developing a RAG-based application.

For more details about DocsQA in Sidekicks, read the Sidekicks Document Q&A guide.

Using the Git + GitHub Document Collection​

When creating a new collection, there is time required to crawl the content, create chunks from the it, and then generate and store embeddings. In the interest of time, we have already created a document collection for use in the tutorial.

The collection has two sources: Git docs and GitHub docs.

Run some queries​

In the Fixie Dashboard, go into Documents. This is where you can view all the Document Collections that you and/or your organization have created.

  1. Click on the card for the "Git + GitHub" Collection.
  2. Click on "Query Test".
  3. Try the following queries:
  • what is a repository
  • what is the gh cli
  • how does rerere work

For each query that you run, you will notice that you are getting back a list of chunks. These chunks were created when Fixie crawled the sources for the document collection. Each chunk also has a score that indicates how similar the chunk is to the query (higher scores mean higher similarity).

Hooking up the GitHub Collection to our Sidekick​

  1. Add the following line to your project's index.tsx file:
const FIXIE_CORPUS_ID: string = '286b5a7d-2bcd-483f-aef5-acf157c5aea5';
  1. Serve up these changes and do some queries using the Sidekick interface in the Fixie Dashboard:
npx fixie@latest serve

In Part 3 we are going to do some work on the system message to make thing work even better.

"Kick"-Starters

This section provides some optional, suggested exercises you can do to go deeper with your usage of Sidekicks.

Creating a new Document Collection

Use the Fixie Dashboard to create a new collection based on your company's website and product docs. Or make a collection for an organization you care about. Once you have created the new collection and added sources, you can use the Document Collection ID and replace the one for Git + GitHub in your project.

Query Chunks via the Corpus API

Fixie provides the Corpus API for working with Document Collections (AKA corpora), sources, chunks, and documents. Go check out this API and use it to perform a query instead of using the UI in the Fixie Dashboard as we did above.