Unverified50% confidenceFactExact time
The paper "CoRD: Coupled Rollouts Fill the Blind Spot of On-Policy Distillation" proposes the coupled rollout distillation (CoRD) method, aimed at actively pushing the distribution of training states toward the teacher side without abandoning the online loss.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/6/2026
First Seen
Valid until: 2/4/2027
Sources
Related Entities
Related Claims
UnverifiedThe core challenge of On-Policy Distillation (OPD) is how to extract effective supervisory signals from the teacher model; traditional methods rely on token probabilities or directly aligning hidden states, each with its own limitations: token probabilities may underestimate the teacher's internal knowledge, while directly aligning hidden states forces accommodation to the teacher's feature space, which is especially tricky in cross-architecture transfer.74% similarUnverifiedCoRD's design automatically anneals the replacement strength over training: as the student gradually learns the teacher-preferred tokens, the two distributions converge and the replacements naturally decrease.69% similarUnverifiedOn-policy distillation requires the student to first generate its own prefix trajectory, after which the teacher provides guidance at the states the student actually reaches, so the state distributions are aligned, and it is usually more stable on long-horizon tasks where errors tend to accumulate.67% similarUnverifiedOn-Policy Distillation (OPD) has become a mainstream means of alleviating the exposure bias of student models.66% similarUnverifiedThe training data for on-policy distillation is generated by real-time sampling of the student model under its current parameters, rather than being a pre-prepared fixed corpus, which contrasts with offline distillation.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1141696API
curl https://kongchang.com/api/v1/knowledge/claims/1141696MCP
get_claim(id=1141696)