Open Models Almanac
Methodspost-trainingsupervised fine-tuning

single mention · post-training

Loss masking in supervised fine-tuning

Masks selected targets from the supervised fine-tuning loss; the specific target-selection rule is not given here.

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

Also called loss masking.

Filed under post-training › supervised fine-tuning