AI & ML · Concept
Alignment
Direct preference optimization (DPO) fine-tuning uses response preferences to change model behavior. A common workflow runs supervised fine-tuning first, then applies DPO to the resulting model.
Last updated
Syntax
client.fine_tuning.jobs.create(model=..., training_file=..., method={"type":"dpo", "dpo": {"hyperparameters": {"beta": ...}}})In-depth API docs are only for subscribers: parameters, examples, common mistakes and interview practice.
SubscribeRelated
Supervised fine-tuningFine-tuning jobsSafety checksSafety categories