Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Validity-Calibrated Reasoning Distillation

About

Reasoning distillation aims to transfer multi-step reasoning capabilities from large language models to smaller, more efficient ones. While recent methods have shown promising gains, they typically rely on static teacher-student hierarchies and frame distillation as trajectory imitation. This is misaligned with the structure of reasoning, where intermediate steps are often locally under-specified: global correctness constrains the final answer, but does not uniquely determine each intermediate move. We propose validity-calibrated reasoning distillation, a framework that treats reasoning distillation as a problem of local learning-signal allocation rather than path alignment. Instead of enforcing token-level imitation, we compare the student's and teacher's proposed next-step actions under the same prefix and use their relative local validity to modulate the strength of the distillation update. This yields a dynamic, context-dependent supervision mechanism that preserves the teacher's structural guidance while adapting update strength to local reasoning quality. Across mathematical reasoning, code generation, and instruction-following benchmarks, our method consistently outperforms strong distillation baselines. These results indicate that effective LLM reasoning distillation is governed not by rigid trajectory imitation, but by principled, locally calibrated allocation of learning signal.

Khouloud Saadi, Di Wang• 2026

Related benchmarks

TaskDatasetResultRank
Code GenerationHumanEval (test)
Pass@144.5
612
Instruction FollowingAlpacaEval
Win Rate76.02
420
Code GenerationMBPP+
Pass@162.7
238
Mathematical ReasoningAMC23
PASS@1 Accuracy35
207
Code GenerationHumanEval+ (test)
Pass@140.9
132
Mathematical ReasoningAIME 24
Pass@1 Accuracy13.3
128
Mathematical ReasoningOlympiad
Pass@1 Accuracy35.4
73
Code GenerationHumanEval+
Pass@164
61
Mathematical ReasoningOlympiad
Pass@124.7
41
Instruction FollowingEvol-Inst
Win Rate57.8
34
Showing 10 of 21 rows

Other info

Follow for update