latent-traits

Tag

Cards List
#latent-traits

Scaling Model-Generated Distillation Data Can Make Latent Teacher Traits More Recoverable

arXiv cs.LG · 2026-08-28 Cached

This paper demonstrates that scaling up off-task model-generated distillation data can amplify latent teacher traits in students, even when the data appears benign, suggesting the need for trait-aware curation in AI training.

0 favorites 0 likes
← Back to home

Submit Feedback