Skip to main content

Databricks: LLM101x Large Language Models: Application through Production

LLMs: Application through Production Syllabus 

PDF Version

Course Description

This course is aimed at developers, data scientists, and engineers looking to build LLM-centric applications with the latest and most popular frameworks. You will use Hugging Face to solve NLP problems, leverage LangChain to perform complex, multi-stage tasks, and deep dive into prompt engineering. You will use data embeddings and vector databases to augment LLM pipelines. Additionally, you will fine-tune LLMs with domain-specific data to improve performance and cost, as well as identify the benefits and drawbacks of proprietary models. You will assess the societal, safety, and ethical considerations of using LLMs. Finally, you will learn how to deploy your models at scale, leveraging LLMOps best practices.

 

By the end of this course, you will have built an end-to-end LLM workflow that is ready for production!

Learning Outcomes

After completing this course, you will be able to: 

  • Apply LLMs to real-world problems in natural language processing using popular libraries, such as Hugging Face and LangChain.
  • Build a custom chat model leveraging open-source LLMs
  • Understand the theory behind foundation models, how to fine-tune foundation models on custom datasets, and the innovations that led to GPT-4 and ChatGPT. 
  • Implement LLMOps and multi-step reasoning best practices.
  • Evaluate the efficacy and bias of LLMs using different methods.

Course Content and Activities

Module

Lessons

Assignments (Verified ONLY)

Introduction

Introduction by Matei Zaharia

Primer on NLP

Language Models

Tokenization

Word Embeddings

Summary

Notebook Demo

Quiz (ungraded)

1 - Applications with LLMs 

Module Overview

Hugging Face

Model Selection

NLP Tasks

Prompts

Prompt Engineering

Summary

Notebook Demo

Quiz #1

Lab #1

2 - Embeddings, Vector Databases, and Search

Module Overview

How does vector search work?

Filtering

Vector Stores

Best Practices

Summary

Notebook Demo

Quiz #2

Lab #2

3 - Multi-stage Reasoning 

Module Overview

Prompt Engineering

LLM Chains

Agents

Summary

Notebook Demo

Quiz #3

Lab #3

4 -  Fine-tuning and Evaluating LLMs

Module Overview

Applying Foundation LLMs

Fine-Tuning: Few-shot learning

Fine-Tuning: Instruction-following LLMs

Fine-Tuning: LLMs-as-a-service

Fine-Tuning: DIY

Evaluating LLMs

Task-specific Evaluation

Guest Lecture from Harrison Chase

Summary

Notebook Demo

Quiz #4

Lab #4

5 - Society and LLMs

Module Overview

Risks and Limitations

Hallucination

Mitigation Strategies

Summary

Notebook Demo

Quiz #5

Lab #5

6 - LLMOps

Module Overview

Traditional MLOps

LLMOps

LLMOps Details

Summary

Notebook Demo

Quiz #6

 

Prerequisites

  • Intermediate-level experience with Python
  • Working knowledge of machine learning and deep learning is helpful

Grading

  • 40% Quizzes (6 graded in total)
  • 60% Labs (5 in total)

Estimated Effort

  • 4-12 hours/week, 6 weeks total

Languages

Content: English   |   Videos: English   |   Transcripts: English

Enrollment Tracks

  • Audit - Freely experience the course during the preview period.
  • Verified - Receive a verified certificate by passing the course with a final grade at or above 70%. 
  • Cost: $99 (US)

 Accessibility Support