Advanced Guide to ChatGPT, Embeddings, and Other Large Language Models (LLMs)

Course details

Looking to learn more about the many potential use cases of large language models (LLMs)? This course is a quick-start guide designed to help you learn how to use and launch ChatGPT, T5, and BERT at scale. With real-world case studies to illustrate the concepts, instructor Sinan Ozdemir outlines a step-by-step approach to building and deploying LLMs with ease. From creating a recommendation engine and launching an information retrieval system to building an image captioning system and beyond, this course provides clear instructions and best practices for anyone interested in using LLMs to generate insights that otherwise would be difficult to obtain.

Learning objectives
Launch an application using proprietary models with OpenAI embeddings and GPT-3.
Fine-tune GPT-3 with custom examples using its API to get better results.
Learn the basics of prompt engineering by building and customizing a chatbot.
Deploy custom LLMs to the cloud.

Concepts

0. Introduction

01 - Quick guide to large language models - Introduction

1. Overview of Large Language Models (LLMs)

02 - Module 1 - Introduction to large language models
03 - Topics
04 - What are language models
05 - Popular modern LLMs
06 - Applications of LLMs

2. Semantic Search with LLMs

07 - Topics
08 - Introduction to semantic search
09 - Building a semantic search system
10 - Optimizing semantic search with cross-encoders and fine-tuning

3. First Steps with Prompt Engineering

11 - Topics
12 - Introduction to prompt engineering
13 - Working with prompts across models
14 - Building a retrieval-augmented generation bot with ChatGPT and GPT-4o

4. Retrieval-Augmented Generation and AI Agents

15 - Topics
16 - Introduction to retrieval-augmented generation (RAG)
17 - Building a RAG bot
18 - Using open-source models with RAG
19 - Expanding into AI agents

5. Optimizing LLMs with Fine-Tuning

20 - Module 2 - Getting the most out of LLMs introduction
21 - Topics
22 - Transfer learning - A primer
23 - The OpenAI fine-tuning API
24 - Case Study - Predicting with Amazon Reviews - Part 1
25 - Case Study - Predicting with Amazon Reviews - Part 2

6. Advanced Prompt Engineering

26 - Topics
27 - Input output validation
28 - Batch prompting and prompt chaining
29 - Chain-of-thought prompting
30 - Preventing prompt injection attacks
31 - Assessing an LLM's encoded knowledge level

7. Customizing Embeddings and Model Architectures

32 - Topics
33 - Case study - Building an anime recommendation system
34 - Using OpenAI s embedding models
35 - Fine-tuning an embedding model to capture user behavior

8. AI Alignment - First Principles

36 - Topics
37 - Introduction to AI alignment
38 - Evaluating alignment plus ethics

9. Moving Beyond Foundation Models

39 - Module 3 - Advanced LLM usage introduction
40 - Topics
41 - The vision transformer
42 - Using cross attention to mix data modalities
43 - Case study - Visual QA Setting up a model
44 - Case study - Visual QA Setting up parameters and data
45 - Introduction to reinforcement learning from feedback
46 - Aligning FLAN-T5 with reinforcement learning from feedback

10. Advanced Open-Source LLM Fine-Tuning

47 - Topics
48 - BERT for multilabel classification - Part 1
49 - BERT for multilabel classification - Part 2
50 - Writing LaTeX with GPT-2
51 - Case study - Sinan s attempt at wise yet engaging responses SAWYER
52 - Instruction alignment of LLMs - Supervised fine-tuning
53 - Instruction alignment of LLMs - Reward modeling
54 - Instruction alignment of LLMs - RLHF
55 - Instruction alignment of LLMs - Using an instruction-aligned LLM

11. Moving LLMs into Production

56 - Topics
57 - Cost projecting and deploying LLMs to production
58 - Knowledge distillation

12. LLM Evaluations

59 - Topics
60 - Evaluating generative tasks - Part 1
61 - Evaluating generative tasks - Part 2
62 - Evaluating understanding tasks
63 - Probing LLMs for world model

Conclusion

64 - Quick quide to large language models - Summary