DA670: RAG Framework Evaluation and Optimization

Upgrade your skills in Retrieval-Augmented Generation (RAG) as you learn to evaluate and optimize real-world retrieval systems. Through hands-on experimentation with cutting-edge VoyageAI embedding and reranking models, you'll master the techniques that drive higher accuracy and smarter search.

Advanced Training: DA670 - RAG Framework Evaluation and Optimization

Overview

This advanced workshop covered evaluating RAG systems using industry-standard metrics and optimization by determining which potential processes within the RAG framework help our application perform better. 

Understanding these concepts enables you to improve your own RAG frameworks and determine which modifications should be made to your systems. 

Intended Audience

This training workshop is intended for developers who have completed the Developer Foundation courses and are using MongoDB Atlas.

      Workshop Objectives

      In this training workshop, you will cover the following topics:

      • Evaluating RAG 

      • Comparing search methods

      • Selecting embedding models

      • Chunking methods

      • Vector quantization

      • Matryoshka embeddings

      • Reranking models

      • Query Rewriting

      • MultiQuery Retrieval

      Duration

      • 1 Day

      Learning Paths

      Please see the MongoDB Learning Paths here!


      Program Datasheets


      Meetings

      Course prerequisites

      This course has the following prerequisites:

      • MDB100: MongoDB Database and Security
      • MDB200: MongoDB Optimization and Performance
      • MDB300: MongoDB Production Readiness
      • DEV400: MongoDB Developer Extension

      Language

      This course is offered in multiple languages. Coordinate with your MongoDB representative to find out more.

      Course access

      This course is entirely virtual and will be conducted over Zoom video conferencing and Instruqt hands-on lab environments.


      View All Upcoming Public Training Sessions

      You can browse all upcoming MongoDB Public Training Sessions sessions via the Instructor-Led Training Schedule.

      How to Enroll in a Live Public Training Session

      • From the course page, select the preferred session from the dropdown menu on the right-hand toolbar.

      • Once the preferred session is selected, click the “Register” button.

        • Note: The session title includes the language, time zone and date of the training.

      How to Switch to a Different Session

      If you’ve already enrolled but want to attend another session of the same course:

      • Go to your Learner Dashboard.

      • Locate the course under “In Progress” or “Events”.

      • Click the three dots to the right of the session title.

      • Select “Re-Enroll” and choose your new preferred session from the dropdown list.

      How to Deregister from a Session

      Currently, learners cannot self-unregister from a live session.

      • To cancel your enrollment, email training@mongodb.com and we will manually remove you.

      Important Information

      See the Learner Guide for more information.

      By registering for a live training session, you confirm your intention to attend. If you miss a session without notifying us, you may be penalized and risk losing access to your training subscription.


      This course is available through Private Training for customers with 7 or more Training Subscriptions.

      Private Training allows you to select specific courses and deliver them to your team in a private classroom, scheduled at your convenience.

      Please contact your Professional Services representative to discuss available dates for private training.

      Important: Private sessions can only be arranged if 7 or more engineers commit to attending.