Unverified50% confidenceFactExact time
Off-policy distillation directly uses trajectories generated by the teacher; the student is exposed to the teacher's distribution, which deviates from its own test-time behavior and can easily cause forgetting or degraded generalization.
1
Sources
50%
Confidence
Long-term
Relevance
11/10/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedThe teacher model outputs in distillation training rarely include self-correction examples, causing student models to form an implicit belief that "strong models don't make mistakes," leading them to repeat rather than correct when verification fails.79% similarUnverifiedOver-reliance on AI for lesson preparation may lead to homogenization of teaching content and weaken teachers' professional judgment.77% similarUnverified许多开源模型继承教师模型的自信偏差,导致工具调用错误时不愿自我纠正75% similarUnverifiedStudents lacking metacognitive ability are more likely to fall into the trap of cognitive offloading, which undermines their own learning capacity.75% similarUnverifiedSimple token removal can only weaken the transfer effect of the teacher's safety tendencies to the student and cannot completely eliminate it; post-hoc DPO alignment can also only partially correct this imprinted behavioral tendency.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1155426API
curl https://kongchang.com/api/v1/knowledge/claims/1155426MCP
get_claim(id=1155426)