AI Concept Atlas
FIELD NOTES / 01
Side by side

Fine-tuning & RAG

RAG supplies retrieved information for generation; fine-tuning changes trainable model parameters. A system can use both.

This is not an either-or choice: a fine-tuned model may still use retrieval. The original RAG research also includes training.

Swap columns ⇄
DimensionFine-tuningRAG
PurposeAdapt an existing model through further training.Bring external evidence into an answer. [Lewis et al.][Hugging Face]
Role in the systemA model adaptation process.A retrieval-and-generation arrangement. [Lewis et al.][Hugging Face]
Inputs and outputsTraining examples and an existing model produce adapted parameters.A query and external collection inform a generated response. [Lewis et al.][Hugging Face]
PersistenceThe adapted parameters remain available after training.The external collection can change separately from model weights. [Lewis et al.][Hugging Face]
ExampleTrain on support examples to improve response behavior.Retrieve the current product manual before answering. [Lewis et al.][Hugging Face]
Boundary caveatsTraining does not automatically provide current facts from an external collection.Retrieved material and the generated answer still need checking. [Lewis et al.][Hugging Face]

Evidence for this distinction

Last editorial review: 2026-09-13.