Advanced Retrieval Augmented Generation

Advanced Retrieval Augmented Generation, available at $54.99, with 48 lectures, and has 1 subscribers.

You will learn about You will learn how to increase the robustness of you LLM calls by implementing structured outputs, acing, caching and retries How to generate synthetic data to establish a baseline for your RAG system, even if your RAG system don't have users yet How to filter out redundant generated data How to make all your LLM calls faster AND cheaper using asynchronous Python and caching How to not be held back by OpenAI rate limits This course is ideal for individuals who are Software Engineer with at least 2 years of Experience or Beginners can follow the video but won't be able to replicate the practical part (They can still learn a lot) or Data Scientists / Analysts with at least 2 years of Experience It is particularly useful for Software Engineer with at least 2 years of Experience or Beginners can follow the video but won't be able to replicate the practical part (They can still learn a lot) or Data Scientists / Analysts with at least 2 years of Experience.

Enroll now: Advanced Retrieval Augmented Generation

Summary

Title: Advanced Retrieval Augmented Generation

Price: $54.99

Number of Lectures: 48

Number of Published Lectures: 45

Number of Curriculum Items: 48

Number of Published Curriculum Objects: 45

Original Price: $199.99

Quality Status: approved

Status: Live

What You Will Learn

You will learn how to increase the robustness of you LLM calls by implementing structured outputs, acing, caching and retries
How to generate synthetic data to establish a baseline for your RAG system, even if your RAG system don't have users yet
How to filter out redundant generated data
How to make all your LLM calls faster AND cheaper using asynchronous Python and caching
How to not be held back by OpenAI rate limits

Who Should Attend

Software Engineer with at least 2 years of Experience
Beginners can follow the video but won't be able to replicate the practical part (They can still learn a lot)
Data Scientists / Analysts with at least 2 years of Experience

Target Audiences

Software Engineer with at least 2 years of Experience
Beginners can follow the video but won't be able to replicate the practical part (They can still learn a lot)
Data Scientists / Analysts with at least 2 years of Experience

Master Advanced Retrieval Augmented Generation (RAG) with Generative AI & LLM

Unlock the Power of Advanced RAG Techniques for Robust, Efficient, and Scalable AI Systems

Course Overview:

Dive deep into the cutting-edge world of Retrieval Augmented Generation (RAG) with this comprehensive course, meticulously designed to equip you with the skills to enhance your Large Language Model (LLM) implementations. Whether you’re looking to optimize your LLM calls, generate synthetic datasets, or overcome common challenges like rate limits and redundant data, this course has you covered.

What You’ll Learn:

Implement structured outputs to enhance the robustness of your LLM calls.
Master asynchronous Python to make your LLM calls faster and more cost-effective.
Generate synthetic data to establish a strong baseline for your RAG system, even without active users.
Filter out redundant generated data to improve system efficiency.
Overcome OpenAI rate limits by leveraging caching, tracing, and retry mechanisms.
Combine caching, tracing, and retrying techniques for optimal performance.
Secure your API keys and streamline your development process using best practices.
Apply advanced agentic patterns to build resilient and adaptive AI systems.

Course Content:

Introduction to RAG and Structured Outputs: Gain a solid foundation in RAG concepts and learn the importance of structured outputs for agentic patterns.
Setup and Configuration: Step-by-step guidance on setting up your development environment with Docker, Python, and essential tools.
Asynchronous Execution & Caching: Learn to execute multiple LLM calls concurrently and implement caching strategies to save time and resources.
Synthetic Data Generation: Create high-quality synthetic datasets to simulate real-world scenarios and refine your RAG system.
Advanced Troubleshooting: Master debugging techniques for async code and handle complex challenges like OpenAI rate limits.

Requirements:

A modern laptop with Python installed or access to Google Drive.
Experience as a software engineer (2+ years preferred).
Intermediate Python programming skills or ability to learn quickly.
Basic understanding of data science (precision, recall, pandas).
Access to a pro version of ChatGPT or equivalent LLM tools.

Who Should Enroll:

Software engineers with experience in basic RAG implementations who want to advance their skills.
Data scientists and AI professionals looking to optimize their LLM-based systems.
Developers interested in mastering the latest RAG techniques for robust, scalable AI solutions.

Join this course today and transform your AI systems with the latest Advanced RAG techniques!

Course Curriculum

Chapter 1: Introduction & Setup

Lecture 1: Introduction

Lecture 2: How To Follow This Course

Lecture 3: Create an OpenAI API Key

Lecture 4: Setup Vitual Env and Jupyter

Lecture 5: Setup local Langfuse with Docker

Lecture 6: How to get the Python `.gitignore` for the next video

Lecture 7: Initialize a git repo for our code

Chapter 2: Section 1 – Making our LLM powered systems more Robust

Lecture 1: Securing our API Keys with python-dotenv

Lecture 2: Calling OPENAI

Lecture 3: [OPTIONAL] Theory : What are Tokens

Lecture 4: [OPTIONAL] Theory: What are LLM, Instructed Models and Chat Templates

Lecture 5: Use asynchronous code to execute several OpenAI LLM calls concurrently

Lecture 6: Introducing the five problems we need to solve to make our LLM apps more robust

Lecture 7: Caching (part 1): How to cache variables on disk

Lecture 8: Caching (part 2): Caching LLM Calls

Lecture 9: [CODE] Source Code for Caching LLM Call

Lecture 10: [Optional] Ignore cache directory in git

Lecture 11: [Optional] Make our cached function work with autocompletion

Lecture 12: How to use Tracing to inspect your systems more easily and build gold datatasets

Lecture 13: Tracing + Caching

Lecture 14: Why Retrying matters and how to implement it with the Backoff Python Decorator

Lecture 15: How to combine Caching, Tracing and Retrying, and a demo of RateLimiteErrors

Lecture 16: [OPTIONAL THEORY] Why are Structured Outputs so important for Agentic Patterns

Lecture 17: Structured Outputs Demo

Lecture 18: Robust "Structured Outputs" Calls with Tracing, Retrying and Caching

Lecture 19: Source Code for Robust "Structured Outputs"

Lecture 20: [OPTIONAL] Section Conclusion

Chapter 3: Section 2 – Measure and Improve Performance of the Retrieval System

Lecture 1: Section Introduction – What is RAG, Retrieval and what's the plan here

Chapter 4: Section 2 – Part 1 – Generating a Synthetic Evaluation Dataset

Lecture 1: Dataset Introduction & How to Download It (it's in the lecture's resources)

Lecture 2: Loading the Dataset & Downloading the Starter Notebook In the Resources

Lecture 3: How are we going to generate a synthetic dataset

Lecture 4: How to write and then iterate on the first draft of a prompt

Lecture 5: Use Jinja and dedent to create robust and simple Prompt Templating

Lecture 6: [Optional Exercise] Improve the First Prompt Draft

Lecture 7: Improve the prompt to improve the quality of our generated questions

Lecture 8: Create a batch workflow to iterate faster

Lecture 9: Finish improving the prompt for our evaluation question generation step

Lecture 10: Run the prompt against the full dataset and display a progress bar to monitor

Lecture 11: Introduction : Deduplicating synthetic data with embeddings

Lecture 12: Introduction to Embeddings (Semantic Vectors) and how to use for deduplication

Lecture 13: Making Robust Embedding Calls to OpenAI

Lecture 14: Computing Cosine Similarity between Two Embeddings

Lecture 15: Download The Starter Notebook

Lecture 16: Compute Embeddings for Each Questions

Chapter 5: Course Recording In Progress

Lecture 1: The course is currently in process of being recorded and uploaded

Instructors

Rémi Connesson | Python – Data Science – Machine Learning – Deep Learning
Data Scientist Freelance & Instructeur

Rating Distribution

1 stars: 0 votes
2 stars: 0 votes
3 stars: 0 votes
4 stars: 0 votes
5 stars: 0 votes

Frequently Asked Questions

How long do I have access to the course materials?

You can view and review the lecture materials indefinitely, like an on-demand channel.

Can I take my courses with me wherever I go?

Definitely! If you have an internet connection, courses on Udemy are available on any device at any time. If you don’t have an internet connection, some instructors also let their students download course lectures. That’s up to the instructor though, so make sure you get on their good side!

Menu