SOURCE-LINKED INTELLIGENCE
Teacher Should Think Ahead: Adaptive Continuations for Reliable On-Policy Distillation
On-policy distillation (OPD) is a promising approach for transferring knowledge between language models, where a student receives dense token-level supervision along its own generated trajectories. However, teacher supervision can be unreliable when conditioned on incomplete or low-quality student prefixes. We identify Teacher Uncertainty Contraction (TUC), a systematic phenomenon whereby the teacher's predictive uncertainty decreases as it continues from a student-generated prefix. We theoretically characterize this trade-off through a variance-bias decomposition of teacher-branch gradients,
Read original source ↗ Open in workspace
- recordType
- paper
- region
- Global
Evidence & attribution
- arXiv · AI, language, vision and robotics · 2026-09-06T07:08:41.000Z
First collected: 2026-09-25T16:52:32.424Z. This is not the publication date.