Open Models Almanac
Methodspost-trainingsupervised fine-tuning

single mention · post-training

Depth-token noise during fine-tuning

Adds noise to depth tokens during fine-tuning as part of a depth-aware setup.

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 10% depth-token noise.

Filed under post-training › supervised fine-tuning