MODEL ADAPTATION • 222 / 397
Know when to prompt, retrieve, fine-tune or compress a model.

Supervised Fine-Tuning (SFT)

SFT fine-tunes a model on input-output examples showing the desired response.

Think of it like

Think of Supervised Fine-Tuning (SFT) as changing how a trained worker behaves or how efficiently that worker can be deployed, rather than giving it a live reference book.

Real life

Organizations adapt general models using techniques such as Supervised Fine-Tuning (SFT).

SRE lens

Teach a model the organization’s incident-summary format using curated examples.

Remember thisSFT fine-tunes a model on input-output examples showing the desired response.
AIForSREJump to a concept
Search is optional. The main journey is simply ↓ Next.