A distribution-correction framework for offline LLM reasoning distillation improves accuracy on math benchmarks by adaptively aligning teacher supervision with the student's inference-time distribution.
Sutton, David McAllester, Satinder Singh, and Yishay Mansour
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
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Distribution Corrected Offline Data Distillation for Large Language Models
A distribution-correction framework for offline LLM reasoning distillation improves accuracy on math benchmarks by adaptively aligning teacher supervision with the student's inference-time distribution.