schema · evergreen

RAG or fine-tune?

RAG is for knowledge, fine-tuning is for form. A decision table and a decision tree.

The shortest split: RAG is for knowledge, fine-tuning is for form. Baking facts that change weekly into weights means retraining every week.

RAGFine-tune
Knowledge livesoutside the modelin the weights
Updating iteasy — reindexhard — retrain
Inference costhigher, context is fulllower, prompt is short
Context limityes, the windownone
Citing sourcesnaturalimpossible
Good atfacts, fresh datatone, schema, behaviour
Rendering diagram…
#rag #fine-tuning #decision

See this note on the whiteboard →