Interacting directly with LLM APIs provides fine-grained control, but building more complex applications often involves repetitive code patterns and managing application state, such as conversation history. LLM frameworks simplify this process by providing abstractions and tools to structure development.
This chapter introduces common LLM application development frameworks, focusing primarily on LangChain as a representative example. You will learn about the core building blocks these frameworks offer, including interfaces for models, prompt templates, and output parsers. We will cover how to link these components together using chains, manage conversational memory, and build basic agents capable of using external tools to perform tasks. The goal is to equip you with the ability to construct more sophisticated and maintainable LLM applications efficiently. Practical exercises will involve building applications using these framework components.
5.1 Introduction to LLM Frameworks (e.g., LangChain)
5.2 Core Components: Models, Prompts, Parsers
5.3 Understanding Chains
5.4 Managing Memory in LLM Applications
5.5 Introduction to Agents
5.6 Using Tools with Agents
5.7 Hands-on practical: Develop an Agentic Application
© 2025 ApX Machine Learning