top of page

Mastering Langchain for LLM Application Development

Time and Location

Thursday, Sept 7, 2023 at 9:00am - 3:00pm EDT

Crowne Plaza Atlanta Perimeter at Ravinia

Agenda

Description

This in-depth workshop offers a comprehensive exploration of LangChain with Python. You will not only gain proficiency in utilizing LangChain but also develop a fully functional generative AI application using this powerful platform.


You'll dive into Prompt Engineering, where you will explore concepts like Chain of Thought, ReAct, and Few-Shot prompting, gaining a deep understanding of LangChain's underlying construction. You'll grasp the various components of LangChain, including Chains, Agents, DocumentLoader, TextSplitter, OutputParser, and Memory, equipping yourself with the knowledge to leverage these tools effectively. Additionally, you'll familiarize yourself with vector stores/vector databases like Chromadb and FAISS. Applying your newfound knowledge, you'll embark on a hands-on project, where you'll develop a custom question-answering application using LangChain and Streamlit.


This workshop provides a concise, balanced learning experience with theory, application, and practice. By its conclusion, you'll have a robust understanding of LangChain's capabilities and applications.

Learning Outcomes

Upon successful completion of this workshop, participants will be able to:

  • Gain proficiency in utilizing LangChain.

  • Achieve the ability to create a fully functional generative AI application using LangChain.

  • Comprehend the theory of Prompt Engineering, including concepts such as Chain of Thought, ReAct, and Few-Shot prompting, as well as understanding the underlying construction of LangChain.

  • Grasp the various components of LangChain, including Chains, Agents, DocumentLoader, TextSplitter, OutputParser, and Memory.

  • Familiarize yourself with vector stores/ vector databases such as Chromadb and FAISS.

  • Apply their learning in a hands-on project, developing a custom question-answering application using LangChain and Streamlit.

Workshop Outline

 Getting Started with LangChain 

  • What is LangChain?

  • Why LangChain?

  • Schema – working with different data types and schemas

  • Text

  • Chat messages

  • Documents

  • Models- interacting with the interface to the AI brains

  • Language model

  • Chat model

  • Text embedding model

  • Prompts – Refining the text for instructions to your model

  • Prompt

  • Prompt template

  • Example selectors

  • Output parsers

  • Indexes - structuring documents for LLMs can work with them

  • Document loaders

  • Text splitters

  • Retrievers

  • Vector stores

  • Memory - storing conversation history and managing context space

  • Chat Message History

  • Chains - predetermined chain of calls to LLMs and other tools.

  • Simple sequential chains

  • Summarization chain

  • Agents – non-deterministic sequence of action to achieve goals

  • Tools

 

LangChain Use Cases 

  • Summarization – summarizing long documents beyond the token limit

  • Question & answering using documents as context

  • Extraction – getting structured data from unstructured text

  • Evaluation – evaluating outputs generated from LLM models

  • Querying tabular data – without using any extra code

  • Code Understanding – co-pilot functionality that can help answer questions from a specific library

  • Interacting with APIs – understand a request from a user and carry out an action

  • Agents with toolkit – run programs autonomously without the need for human input

 

Project : Question-Answering App on your Custom Data 

  • Project introduction

  • Loading your custom (private) pdf documents

  • Loading different document formats

  • Chunking strategies and splitting the documents

  • Embedding and uploading to a vector database (Chromadb)

  • Adding memory (chat history)

  • Creating front-end of application using streamlit (Python framework to build ML webapps)

Pre-requisites

For optimal learning outcomes, participants should meet the following prerequisites:

\

  • Familiarity with Python: As some parts of the workshop involve coding or code interpretation, a basic understanding of Python is required.

  • Curiosity about Large Language Models (LLMs): As LangChain operates primarily around LLMs, a keen interest in learning about different LLM frameworks will enhance the workshop experience and aid in comprehension.

bottom of page