Open Models Almanac
Methodstraining objectiveauxiliary loss

single mention · training objective

Squared log-policy-ratio regularizer

Penalizes policy drift using a squared log-ratio term.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See auxiliary loss for the methods filed beside it.

Also called squared log-ratio term.

Filed under training objective › auxiliary loss