A clever way to sidestep some of AI's biggest issues is on the rise

Companies and developers looking to implement LLMs are increasingly turning to retrieval augmented generation, or RAG, before fine-tuning and pre-training.

Aug 23, 2023

∙ Paid

A robot working in a factory covered with bandaids and duct tape, funny comic aesthetic — midjourney

In an ideal world, every company could figure out how to build some kind of customized AI model to suit their needs—either open source or through a API provider.

But fine-tuning isn’t easy, and it isn’t cheap either, whether that’s cost per token or cost p…

A clever way to sidestep some of AI's biggest issues is on the rise

Companies and developers looking to implement LLMs are increasingly turning to retrieval augmented generation, or RAG, before fine-tuning and pre-training.

This post is for paid subscribers