Retrieval-augmented technology (RAG) is a way used to “ground” massive language fashions (LLMs) with particular knowledge sources, typically sources that weren’t included within the fashions’ authentic coaching. RAG’s three steps are retrieval from a specified supply, augmentation of the immediate with the context retrieved from the supply, after which technology utilizing the mannequin and the augmented immediate.
At one level, RAG appeared like it will be the reply to all the things that’s incorrect with LLMs. While RAG may also help, it isn’t a magical repair. In addition, RAG can introduce its personal points. Finally, as LLMs get higher, including bigger context home windows and higher search integrations, RAG is turning into much less crucial for a lot of use instances.
Meanwhile, a number of new, improved sorts of RAG architectures have been launched. One instance combines RAG with a graph database. The mixture could make the outcomes extra correct and related, significantly when relationships and semantic content material are vital. Another instance, agentic RAG, expands the assets obtainable to the LLM to incorporate instruments and features in addition to exterior data sources, corresponding to textual content databases.
