GraphRAG, a graph-based approach to retrieval-augmented generation (RAG) that significantly improves question-answering over private or previously unseen datasets, is now available on GitHub. Learn more. https://msft.it/6040l8lVO
Knowledge graphs + LLMs = 🔥🔥🔥 This combination, also used in the new HippoRAG paper, is shaping up to be a great multi-purpose RAG framework. Having an LLM create the knowledge graph and then using the graph for retrieval is just 🤌
For question answering, factual questions must be already seen or trained with LLM. for temporal questions answering, if structured info can be extracted from new dataset and build a high quality KG, it could be directly used for answer retrieval. is there a benefit on building the context and feed into a LLM for answer generation? 🤔
We are looking forward to implementing this on our app.
We’ve been using graphs for a while, it’s a great approach for complex queries
📌📌My Takeaway: - GraphRAG improves question-answering on unseen datasets - Available on GitHub for exploration 🚀 PS: Let's RAG and roll!
Can individual KG Communities be containerized while grooving shortest paths between two different and distinct communities? 🙏 #GraphRAG
Muhammad Daniyal can you add this to our exel list please
Let's see if it was worth the wait.
looking forward to test this out
Co-Founder, Board Member at Aimped | AI Solutions Architect | Generative AI | NLP Certification, Master's Degree | Taught 3000+ ML Students Globally | Digital Marketing
3wThis is amazing news that we have been eagerly anticipating for a while at Aimped AI