Skip to content

ExamplesΒΆ

  • Lazy Graph RAG

    Implements LazyGraphRAG using LangChain and langchain-graph-retriever.

    It loads Wikipedia articles and traverses based on links ("mentions") and named entities (extracted from the content). It retrieves a large number of articles, groups them by community, and extracts claims from each community. The best claims are used to answer the question.

    Lazy Graph RAG Example

  • Code Generation

    This example notebook shows how to load documentation for python packages into a vector store so that it can be used to provide context to an LLM for code generation.

    It uses LangChain and langchain-graph-retriever with a custom traversal Strategy in order to improve LLM generated code output. It shows that using GraphRAG can provide a significant increase in quality over using either an LLM alone or standard RAG.

    GraphRAG traverses cross references in the documentation like a software engineer would, in order to determine how to solve a coding problem.

    Code Generation Example