Edge 303: The Top Two Types Retrieval-Augmented Language Models
What are the main types of techniques to augment LLMs with external information.
In this Issue:
The two main types of retrieval augmented LLMs.
Meta AI’s Atlas retrieval augmented LLM.
The Lamini framework for fine-tuning LLMs.
💡 ML Concept of the Day: Two Types Retrieval-Augmented Language Models
In the previous edition of this series, we introduced the concept of retrieval augmented LLMs(RALMs) as models that can complement its knowledge by accessing information from external sources. The RALM research was pioneered by Google a few years ago and has rapidly become one of the fundamental elements of modern LLMs. Not all RALM techniques are equal and we can divide the space into two fundamental groups of methods: