Aussie AI

Retrieval Augmented Language Models (RALM)

  • Last Updated 2 March, 2025
  • by David Spuler, Ph.D.

Retrieval Augmented Language Models (RALM) is the general method of using external data sources to make LLMs more powerful. It improves the "smartness" of the LLM, rather than being a speed optimization. In fact, it is often slower, because accessing a secondary data source requires an extra step.

Types of RALM include:

Also related in the sense of providing extra context to the LLM, but not technically part of RALM are techniques including:

RALM vs RAG

RALM and RAG are almost the same thing, but RALM is a little more general in the types of extra data used as context. RAG is a very specific type of RALM, which a particular architecture, whereas RALM is the general idea.

RALM may also includes capabilities such as:

  • Data source integrations ("plug-ins")
  • Tool Augmented Language Models (TALM); see also tool usage by LLMs.

RALM generally refers to a read-only type architecture that simply returns information for the LLM to use as context, whereas more powerful two-way integrations with tools that "do" something are called "agent architectures."

Research Papers on RALM

Papers on the use of RALM techniques in LLMs and Transformer architectures:

More AI Research

Read more about: