دوره جامع ChatGPT و LangChain برای توسعه‌دهندگان

How to Get Help

What is LangChain

How a Typical AI-Enabled App Works

Here It Is, This is Why We Use LangChain

Project Overview and Setup

Using LangChain the Simple Way

Introducing Chains

Adding a Chain

Parsing Command Line Arguments

Securing the API Key

Connecting Chains Together

Chains in Series with SequentialChain

App Overview

Receiving User Input

Chat vs Completion Style Models

Representing Messages with ChatPromptTemplates

Implementing a Chat Chain

Understanding Memory

Using ChatBufferMemory to Store Conversations

Saving and Extending Conversations

Summarizations Conversation Summary Memory

Project Overview

Project Setup

Loading Files with Document Loaders

Search Criteria

Introducing Embeddings

The Entire Embedding Flow

Chunking Text

Generating Embeddings

Introducing ChromaDB

Building a Retrieval Chain

What is a Retriever

Optional Understanding Refine, MapReduce, and MapRerank

Removing Duplicate Documents

Creating a Custom Retriever

Custom Retriever in Action

Visualizing Embeddings

App Overview

Understanding Tools

Understanding ChatGPT Functions

Defining a Tool

Defining an Agent and AgentExecutor

Understanding Agents and AgentExecutors

Shortcomings in ChatGPT's Assumptions

Recovering from Errors in Tools

Adding Table Context

Adding a Table Description Tool

Being Direct with System Messages

Adding Better Descriptions for Tool Arguments

Tools with Multiple Arguments

Memory vs Agent Scratchpad

Preserving Messages with Agent Executor

Understanding Callbacks

Implementing a Basic Callback Handler

More Handler Implementaion

App Overview

Taking a Look at Mockups

Boilerplate Setup

How This App is Designed

Outlining the First Feature

Loading and Splitting From a PDF

Testing the PDF Upload

Introducing Pinecone

Initializing the Pinecone Client

Adding Documents to the Vector Store

Why is Processing Taking Forever

Introducing Background Jobs

Redis Setup

Adding in the Worker

Queuing Up Jobs

Updating Document Metadata

Understanding the Apps Requirements

Persistent Message Storage

Introducing the Conversational Retrieval Chain

Building the Retriever

Custom History Objects

Building a Custom SQL History

Testing the Chain

Streaming Text Generation

Creating a Working Playground

Experimenting with a Streaming Language Model

Chains Don't Want to Stream

Receiving Chunks with a Callback

Extending a LLM Chain

Adding a Queue for Communication

The Chain Really Wants to Wait

Solving the Slow Chain

It Works!

Ending the Loop

Isolating the Queue and Handler

Using a Mixin Approach

Integrating the Streaming Code

Testing the Streaming Setup

Here's the Issue

Isolating the Handler

Streaming Complete!

Random Component Parts

Component Part Flow

Partial KWArg Application

Building Component Maps

Randomly Picking a Component

Generalizing Component Picking

Collecting User Feedback

Redis Connection Setup

Storing Votes in Redis

Weighted Randomness

Extracting Scores

Calculating the Average Score

Selecting Components By Score

Adding Score Observability

Building the Score Aggregate

Adding Another Form of Memory

Window Memory Implementation

Text Generation Tracing

Langfuse Signup

Adding in Tracing

Understanding the Trace

Automatic Trace Creation