Open Models Almanac
Methodstraining objectiveauxiliary loss

single mention · training objective

Depth-prefix token replacement noise during fine-tuning

Replaces a fraction of depth-code input tokens with uniformly sampled codes while leaving prediction targets unchanged.

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