Open Models Almanac
Methodstraining objectiveauxiliary loss

single source · training objective

Disable auxiliary and hard-target losses during QAD

Disables specified hard-target, MTP, MoE load-balancing, and intermediate-feature losses during quantization-aware distillation.

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.

Filed under training objective › auxiliary loss