When would you choose to use Retrieval-Augmented Generation (RAG) versus Fine-Tuning for an LLM application?

Question

Anonymous · Accepted Answer

I'd prioritize rag when the use case involve (1) searching the content within a
large database of documents and (2) requiring to cite the exact part of the
document as evidence for answering the questions.

For fine-tuning an LLM application, it'd be useful for guiding the model to
improve its performance on unknown tasks based on known knowledge; For instance,
with one-shot approach, we can provide some examples to guide the model avoid
loopholes and increase its performance based on predefined knowledge.

When would you choose to use Retrieval-Augmented Generation (RAG) versus Fine-Tuning for an LLM application?

Practice this question with AI

Go Premium

Community Answers

Unlock Full Access

Practice More Questions