Unverified50% confidenceFactExact time
The HCN (Highest Coverage Next) strategy dynamically selects teachers based on their weighted coverage of the student's unsolved problems.
1
Sources
50%
Confidence
Long-term
Relevance
11/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTeacher intervention pilot studies show that teacher preferences differ across tokens, and that through majority voting and KL-divergence-based selection, a better teacher can be identified for each token.67% similarUnverifiedIn on-policy distillation, the student model generates its own rankings during training and the teacher provides feedback on these rankings, thereby avoiding the mismatch between teacher and student distributions (exposure bias) in offline distillation.65% similarUnverifiedSelf-Distillation Policy Optimization (SDPO) provides dense token-level supervision signals to the student model through a feedback-conditioned "self-teacher".65% similarUnverifiedToken-level supervision means that the teacher gives a distribution or label for each token position in the sequence, and the student learns by minimizing the difference from the teacher's distribution (such as KL divergence).64% similarUnverifiedA joint scaling strategy that scales the critic in tandem with the teacher prevents training collapse and enables the student model to benefit from a larger teacher.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1154302API
curl https://kongchang.com/api/v1/knowledge/claims/1154302MCP
get_claim(id=1154302)