Scribd
Upload
108 views2 pages

LangChain Cheat Sheet

LangChain is a framework designed for building applications using large language models (LLMs), incorporating tools, memory, data, and reasoning capabilities. It features core modules for model interaction, retrieval, chaining, agent actions, memory storage, and callbacks for debugging. Additionally, it supports prompt engineering, output parsing, evaluation, and data connection through RAG pipelines, while providing functionalities for text splitting and saving prompts.

Uploaded by

unirockcreations
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
108 views2 pages

LangChain Cheat Sheet

LangChain is a framework designed for building applications using large language models (LLMs), incorporating tools, memory, data, and reasoning capabilities. It features core modules for model interaction, retrieval, chaining, agent actions, memory storage, and callbacks for debugging. Additionally, it supports prompt engineering, output parsing, evaluation, and data connection through RAG pipelines, while providing functionalities for text splitting and saving prompts.

Uploaded by

unirockcreations
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

LangChain Cheat Sheet

LangChain Summary
LangChain = Framework for LLM apps.

Adds tools, memory, data & reasoning.

Supports OpenAI, Anthropic, etc.

Core Modules
1. Model I/O Interface with LLMs.

2. Retrieval Search docs/knowledge.

3. Chains Prompt model output.

4. Agents LLM decides actions/tools.

5. Memory Stores chat context.

6. Callbacks For logging/debugging.

Chat Message Types


SystemMessage Rules/instructions.

HumanMessage User input.

AIMessage LLM reply.

Prompt Engineering
Use PromptTemplate / ChatPromptTemplate.

Combine with `|` pipe.

Few-shot with examples.

Output Parsing
Use Output Parsers (not regex).

PydanticOutputParser structured data.

Use .get_format_instructions() + .parse().

Evaluation
Use LangChain Evals.

e.g., labeled_pairwise_string for comparing answers.


LangChain Cheat Sheet

Check accuracy, similarity, rankings.

OpenAI Function Calling


LLMs return JSON to trigger external tools.

Connecting to Data (RAG Pipelines)


Load Chunk Embed Store Retrieve.

Use unstructured (PDF, CSV) or structured (SQL) data.

Text Splitting
CharacterTextSplitter by characters.

TokenTextSplitter by tokens (e.g., tiktoken).

Saving Prompts
Save as .json or .yaml with .save().

Reuse with load_prompt().

Query Planning
Use QueryPlan to split complex queries into smart steps.

You might also like