Skip to content

liamjdavis/EDGAR-RAG-Web-App

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EDGAR RAG

EDGAR RAG (Retrieval-Augmented Generation) is a web interface designed to efficiently retrieve and present data from SEC filings, specifically the 10-Q and 10-K forms available through the EDGAR database. The application leverages a RAG fine-tuned on Microsoft Phi-3.5 Mini to provide accurate and contextually relevant information from these financial documents.

Features

  • Advanced Search: Quickly retrieve relevant data from 10-Q and 10-K filings using a powerful search engine backed by a vector database.
  • Contextual Answers: The RAG model generates precise answers based on the context provided in SEC filings, reducing the need to manually sift through documents.
  • User-Friendly Interface: The web interface is designed for ease of use, allowing users to input queries and get results with minimal effort.
  • Citations: Each response has a citation at the end that includes company, form type, and page number.

Architecture

The architecture of the EDGAR RAG application consists of several key components:

  • Frontend: Built using Next.js, the frontend provides a responsive and interactive interface for users to query SEC filings.
  • Backend: The FastAPI backend handles requests from the frontend, interacts with the vector database, and communicates with the RAG model to generate responses.
  • Postgres Database: Handles user authentication and stores previous threads + chats

Demo

Here is a live video demo.

Note on RAG accuracy

The metrics outputed by the RAG are in no way guaranteed to be accurate. Every metric and datapoint should be checked against the citation.