LlamaIndex 🦙 (@llama_index)
2024-06-01 | ❤️ 409 | 🔁 73
Build-your-own Graph RAG 🕸️
There are two prepackaged ways to do RAG with knowledge graphs: vector/keyword search with graph traversal, and text-to-cypher.
Here’s how you can define a custom graph RAG flow, by combining the above two or defining a separate flow entirely 💡
Our latest cookbook shows you how to build a custom graph retriever that combines both vector search and graph search, and performs reranking at the end. This is just one of many ways you can extend graph retrieval for your own use case!
Check it out: https://developers.llamaindex.ai/python/examples/property_graph/property_graph_custom_retriever/
미디어
