Open Models Almanac
Methodstraining objectiveauxiliary loss

single mention · training objective

Attention masking to prevent information leakage between action representations

Masks FAST action tokens from attention to prevent leakage between the two action representations.

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