Course 3 : Generative AI with Large Language Models – AI school

About Course

Course Title: Advanced Topics in Natural Language Processing

Course Description: This advanced course delves into cutting-edge techniques and applications in natural language processing (NLP), focusing on the latest developments in Generative AI and Large Language Models (LLMs). Participants will gain hands-on experience and theoretical knowledge in text generation, prompting techniques, embeddings with OpenAI GPT, question answering, retrieval augmented generation (RAG), preprocessing unstructured data, and leveraging knowledge graphs for RAG applications.

Course Outline:

Exploring Generative AI & LLMs
- Introduction to Generative AI and its applications in NLP
- Overview of Large Language Models (LLMs) and their architecture
- Case studies and examples of state-of-the-art LLMs
Text Generation
- Techniques for text generation using LLMs
- Fine-tuning LLMs for specific text generation tasks
- Generating coherent and diverse text outputs
Techniques in Prompting and Prompt Engineering
- Understanding the importance of prompts in LLMs
- Prompting strategies for controlling and directing LLM outputs
- Advanced prompt engineering techniques
Embeddings with OpenAI GPT
- Introduction to word embeddings and their role in NLP
- Utilizing OpenAI GPT for generating embeddings
- Fine-tuning embeddings for downstream NLP tasks
Question Answering – using embeddings
- Overview of question answering systems
- Leveraging embeddings for question answering tasks
- Case studies and practical applications
Retrieval Augmented Generation (RAG) – Building Applications with Vector Databases
- Understanding retrieval augmented generation (RAG)
- Building RAG systems using vector databases
- Case studies and hands-on projects
Preprocessing Unstructured Data for LLM Applications
- Techniques for preprocessing unstructured data
- Data cleaning, normalization, and tokenization
- Handling different data formats and sources
Knowledge Graphs for RAG
- Introduction to knowledge graphs and their relevance in RAG
- Building knowledge graphs for storing and retrieving information
- Integrating knowledge graphs with RAG systems for enhanced performance

Generative AI Fundamentals
Advanced Text Generation Techniques
Prompt Engineering Strategies
Embeddings and OpenAI GPT
Effective Question Answering
Retrieval Augmented Generation (RAG)
Vector Databases in Applications
Preprocessing Unstructured Data
Leveraging Knowledge Graphs
Practical Application Development

Course Content

Exploring Generative AI & LLMs

List of popular LLMs (OpenAI, Llama, BLOOM…) – key features and comparison
connecting to OpenAI
connecting to Azure OpenAI

Text generation

Techniques in Prompting and Prompt Engineering

Embeddings with OpenAI GPT

Text Embedding Simplified: ADA-002
Embedding with Azure OpenAI API
Amazon food reviews : create embeddings
choosing smaller embedding dimension
Performing semantic search
Visualizing embeddings (t-SNE)
ML with embeddings
Zero shot learning
Question answering using embeddings-based search
Perform search over the embeddings stored in a dataframe

Question Answering – using embeddings

Retrieval Augmented Generation (RAG) – Building Applications with Vector Databases
Master the development of six innovative applications leveraging vector databases: semantic search, retrieval-augmented generation (RAG), anomaly detection, hybrid search, image similarity search, and recommender systems. Each application harnesses a distinct dataset to power its functionalities. Explore the creation of six dynamic applications employing vector databases and deploy them using Pinecone. Craft a hybrid search application amalgamating text and images to enhance multimodal search outcomes. Discover the process of constructing an application that quantifies and prioritizes facial similarity.

fundamentals of information retrieval
concept of retrieval-augmented generation
Evaluation Metrics
Applications and Use Cases
Challenges and Solutions
Advanced Techniques
Hands-on RAG impl – sci claims data
Hands-on RAG impl – oscar awards data
Recommender System
Hybrid Search
Facial Similarity

Preprocessing Unstructured Data for LLM Applications

Knowledge Graphs for RAG
Knowledge graphs serve as indispensable tools in software development, enabling the structuring of intricate data relationships, empowering intelligent search functionalities, and facilitating the creation of robust AI applications capable of reasoning across diverse data types. These graphs possess the unique ability to interlink data originating from both structured and unstructured sources, such as databases and documents, presenting an intuitive and adaptable approach to modeling complex real-world scenarios. Unlike conventional data representations like tables or simplistic lists, knowledge graphs excel in capturing the underlying meaning and contextual nuances of the data they encapsulate. This capability enables users to uncover profound insights and connections that might elude detection when utilizing traditional database structures. The rich, structured context provided by knowledge graphs proves invaluable, particularly in enhancing the output of large language models (LLMs). By leveraging knowledge graphs, developers can furnish LLMs with more pertinent context, surpassing the capabilities offered by semantic search in isolation. In essence, knowledge graphs serve as dynamic frameworks that not only facilitate efficient data organization but also empower AI systems with the contextual understanding necessary to generate more relevant and insightful outputs.

Introduction to Knowledge Graphs
Data Modeling Techniques for Knowledge Graphs
Representation Learning for Knowledge Graphs
Knowledge Graph Construction and Population
Querying and Reasoning in Knowledge Graphs
Applications of Knowledge Graphs in AI and Machine Learning
Knowledge Graph Integration with Natural Language Processing (NLP)
Knowledge Graph Visualization and Exploration
Knowledge Graphs in Semantic Search and Information Retrieval.
Add a vector index to a knowledge graph to represent unstructured text data and find relevant texts using vector similarity search.
Build a knowledge graph of text documents from scratch, using publicly available financial and investment documents as the demo use case
Explore advanced techniques for connecting multiple knowledge graphs and using complex queries for comprehensive data retrieval.
Write advanced Cypher queries to retrieve relevant information from the graph and format it for inclusion in your prompt to an LLM.

Student Ratings & Reviews

No Review Yet

About Course

What Will You Learn?

Course Content

Exploring Generative AI & LLMs

List of popular LLMs (OpenAI, Llama, BLOOM…) – key features and comparison

connecting to OpenAI

connecting to Azure OpenAI

Text generation

Evolution of Text Generation Pre-Transformers

Understanding Transformers Architecture

Text Generation Techniques with Transformers

Techniques in Prompting and Prompt Engineering

The Art of Crafting Prompts: Principles, techniques & best practices

Type of Prompts: Zero shot, One Shot , Few shot prompts, Chain-of-thought etc

Conceptual understanding – Tokens, Max Tokens, temparature

Standard methods for formatting, summarizing, inferring prompts to get best results.

Use OpenAI or AZ OpenAI – Question Anwering, NLI examples

Embeddings with OpenAI GPT

Text Embedding Simplified: ADA-002

Embedding with Azure OpenAI API

Amazon food reviews : create embeddings

choosing smaller embedding dimension

Performing semantic search

Visualizing embeddings (t-SNE)

ML with embeddings

Zero shot learning

Question answering using embeddings-based search

Perform search over the embeddings stored in a dataframe

Question Answering – using embeddings

Question Answering with Chromadb and OpenAI

Build a Q&A Bot for Academy Awards Based on ChatGPT

fundamentals of information retrieval

concept of retrieval-augmented generation

Evaluation Metrics

Applications and Use Cases

Challenges and Solutions

Advanced Techniques

Hands-on RAG impl – sci claims data

Hands-on RAG impl – oscar awards data

Recommender System

Hybrid Search

Facial Similarity

Preprocessing Unstructured Data for LLM Applications

Extract, preprocess, normalize data

Build a RAG (multi doc)

Introduction to Knowledge Graphs

Data Modeling Techniques for Knowledge Graphs

Representation Learning for Knowledge Graphs

Knowledge Graph Construction and Population

Querying and Reasoning in Knowledge Graphs

Applications of Knowledge Graphs in AI and Machine Learning

Knowledge Graph Integration with Natural Language Processing (NLP)

Knowledge Graph Visualization and Exploration

Knowledge Graphs in Semantic Search and Information Retrieval.

Add a vector index to a knowledge graph to represent unstructured text data and find relevant texts using vector similarity search.

Build a knowledge graph of text documents from scratch, using publicly available financial and investment documents as the demo use case

Explore advanced techniques for connecting multiple knowledge graphs and using complex queries for comprehensive data retrieval.

Write advanced Cypher queries to retrieve relevant information from the graph and format it for inclusion in your prompt to an LLM.

Student Ratings & Reviews