Glossary · Model Training

SFT (Supervised Fine-Tuning)

A training stage where a model learns from labeled example input-output pairs, typically used before preference-based alignment techniques like RLHF or DPO.