Example Document GPT

This repository is a Matatika Example Workspace that uses Open AI Generative Pre-trained Transformer (GPT) to answer questions.

Matatika loads the document vectors for the published datasets into Elastic Search and uses the Matatika questions API to generates answers.

Get started

Get started using either Matatika Cloud or locally with Matatika Community Edition.

Setup for local documents

1. Create a new workspace using this workspace as the template existing repository
2. Run the "Load local documents" pipeline to load all document from sample_documents

Setup for remote documents

1. Prerequisite - you will need a folder in Google Drive for this pipeline
2. Prerequisite - with Matatika Community Edition you will need to configure your Google OAuth credentials
3. Create a new workspace using this workspace as the template existing repository
4. Provide configuration for the "Load remote documents" pipeline
   1. Under the gdrive section, login with your Google account and authorize Matatika to access Google Drive
   2. Select the folder with your documents to index
   3. Save the pipeline
5. Run the pipeline

---

Once the pipeline run completes, a dataset will have been created for each loaded document, with a description containing the parsed document content and tags representing the the source file path.

Question Answering

In addition to indexing the published documents for search, this workspace has the questions feature enabled see workspace.yml -> FEATURE_QUESTIONS

Once in the Matatika App or Lab, you can simply research topics and generate answers.

Look out for the big '+' button on any page.