4.77 out of 5
4.77
15 reviews on Udemy

AI Apps with ChatGPT and LangChain: The Introduction

Building Generative AI applications with powerful Large Language Models
Instructor:
Kris Celmer
34 students enrolled
English [Auto]
Understand Large Language Models and how to use them in AI Apps, how to prepare inputs, call GPT API endpoints and process outputs.
Core Prompting Techniques: Instructions, Zero-Shot, Few-Shot, Self Confidence, Chain-of-Thoughts, Output Formatting to JSON Strings, etc.
Using ChatGPT API in Natural Language Processing tasks: Named Entity Extraction, Classification, Sentiment Analysis, Translation.
Manage chat history with LangChain Conversation Memory classes, including Window Buffer, Token Buffer and Summary, enabling sophisticated, intelligent chatbots.
Use Embeddings to create Vector Databases and Semantic Similarity Search to retrieve relevant documents to chat with PDF or HTML docs and Source Code.
Summarise long texts with LangChain Map-Reduce and Refine Chains, including PDFs, web pages, YouTube video transcripts.
Use ReAct and OpenAI Functions Agent Classes to develop reasoning Agents capable of dealing with complex problems, requiring sequence of steps to solve.
Build Chat applications to perform ad hoc Data Analysis with Pandas dataframes or SQL databases.
Develop intuitions about how all these work and can be applied to solve real tasks.
Learn latest and greatest Python libraries to build LLM enabled applications, including the famous LangChain.

ChatGPT revolutionises businesses, how we work and greatly influences our lives. It is much more than a famous Web and mobile applications everyone is using now. Its creators recently released a publicly available API enabling creation of sophisticated AI Apps utilising the power of GPT models to most difficult Natural Language Processing tasks and beyond.

This course aims at Python developers to teach how to harness the power of latest and greatest Large Language Models in custom, innovative applications, how to interface existing data in various formats with ChatGPT available through the API.

You will learn the magic of LangChain – the Python Library delivering ever growing ecosystem of tools and integrations necessary to build the AI Apps. LangChain offers not only convenient wrappers around ChatGPT model APIs, but has plenty of ready-made classes and functions facilitating creation and use of Chat Memory, Vector DBs for semantic search of relevant documents, and blueprints of powerful Agents, capable of using Python functions in your environment to get access to local, proprietary data.

The course is very practical and consists of dozens of practical demonstrations of Python code solving various AI tasks. You will get detailed, precise and in-depth explanation of all presented concepts and algorithms.

All of the code used in the course is available for your download from GitHub repository. You can use it as a basis to further exploration and experimentation leading to quick and easy development of real-life AI Apps.

Course Introduction

1
Introduction
2
Downloading course materials and setting up the environment

Introduction to LLMs

1
Large Language Models
2
ChatGPT - a prime example of LLM
3
ChatGPT for Developers
4
Examples of tasks solved with ChatGPT

Getting started with OpenAI API and LangChain

1
Model variants in OpenAI API
2
Setting up an OpenAI API account
3
Using LangChain library to interact with OpenAI API
4
In-Context Learning and Action Planning design patterns
5
Overview of LangChain concepts

Prompt Engineering for Developers

1
Basic prompt tactics
2
Prompt techniques to improve output quality
3
Prompt Templates
4
Post-processing of ChatGPT responses
5
Introduction to Chains
6
Moderation techniques
7
Example AI App: Analysis of Customer Reviews

Chatbots

1
Closer look at Chat vs QA
2
Chat conversation memory
3
Interactive Chatbot in Jupyter notebook
4
Handling long chat memory

Documents and Web pages

1
Summary of long PDF document
2
Summary with Refine chain
3
Working with Web pages
4
Summarising YouTube video transcript
5
Naive QA with a PDF document

Vector Databases

1
Semantic search and Text embeddings
2
QA over a PDF document with Vector DB
3
QA with Pandas User Guide - Build Persistent Vector Database
4
QA with Pandas User Guide - Use Persistent Vector Database
5
QA with TikToken code
6
QA with Vector DB - Limitations

Action planning agents

1
Smooth intro to Agents
2
ReAct Agent internals
3
OpenAI Functions agent
4
OpenAI Functions agent with chat memory
5
Agent with Vector DB
6
Web search with Bing
7
Action planning and custom tools
8
CSV file with Pandas agent
9
SQL Database agent
10
Managing multiple Tools
Faq Content 1
Faq Content 2

Productivity Hacks to Get More Done in 2018

— 28 February 2017

  1. Facebook News Feed Eradicator (free chrome extension) Stay focused by removing your Facebook newsfeed and replacing it with an inspirational quote. Disable the tool anytime you want to see what friends are up to!
  2. Hide My Inbox (free chrome extension for Gmail) Stay focused by hiding your inbox. Click "show your inbox" at a scheduled time and batch processs everything one go.
  3. Habitica (free mobile + web app) Gamify your to do list. Treat your life like a game and earn gold goins for getting stuff done!


4.8
4.8 out of 5
15 Ratings

Detailed Rating

Stars 5
11
Stars 4
3
Stars 3
1
Stars 2
0
Stars 1
0