Open Models Almanac
Methodstraining objectivemulti-token prediction objective

single mention · training objective

Frozen-backbone MTP-head fine-tuning

An MTP-head training phase in which the backbone is fixed and only the head receives gradient updates.

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

Filed under training objective › multi-token prediction objective