OPD^2 improves multilingual math reasoning over standard distillation, narrows the English-Korean gap, but English-only training shifts output language to English.
Entropy-aware on-policy distillation of language models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
On-Policy Delta Distillation for Multilingual Math Reasoning
OPD^2 improves multilingual math reasoning over standard distillation, narrows the English-Korean gap, but English-only training shifts output language to English.