Critique-Guided Distillation trains models to refine responses using teacher critiques during training only, yielding 7% average gains on math benchmarks and preserving general instruction-following unlike Critique Fine-Tuning.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Critique-Guided Distillation for Robust Reasoning via Refinement
Critique-Guided Distillation trains models to refine responses using teacher critiques during training only, yielding 7% average gains on math benchmarks and preserving general instruction-following unlike Critique Fine-Tuning.