SR-OPSD anchors an evolving self-teacher to a frozen reference policy and projects the student toward the anchored target with Rényi divergence, improving stability and benchmark performance.
Proceedings of the 41st International Conference on Machine Learning , pages =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SR-OPSD: Self-Referenced On-Policy Self-Distillation
SR-OPSD anchors an evolving self-teacher to a frozen reference policy and projects the student toward the anchored target with Rényi divergence, improving stability and benchmark performance.