AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Where Did This Sentence Come From? Tracing Provenance in LLM Reasoning Distillation

kaiyuan liu, Shaotian Yan, Rui Miao, Bing Wang, Chen Shen, Jun Zhang, Jieping Ye

ICLR 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Select responses with most Teacher Sentences vs random selectionQwen3-4B-Base distilled from DeepSeek-R1 · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 56.2 -> 58.7Improved
Select responses with most Teacher Sentences vs random selectionQwen3-8B-Base · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 60.9 -> 62.6Improved
Select responses with most Teacher Sentences vs random selectionQwen2.5-7B-Instruct (QwQ-32B teacher) · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 56.0 -> 57.7Improved
Select responses with most Teacher Sentences vs random selectionQwen3-4B-Instruct-2507 (GPT-OSS-120B teacher) · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 74.0 -> 76.4Improved
Use GRAPE selection instead of randomQwen3-4B-Base · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 56.2 -> 55.1Hurt
Select by relative proportion of Teacher Sentences instead of absolute count (ablation)Qwen3-4B-Base · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 58.7 -> 56.4Hurt
Select responses with fewest Teacher Sentences (ablation)Qwen3-4B-Base · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 58.7 -> 55.0Hurt
Select longest response instead of most Teacher Sentences (ablation)Qwen3-4B-Base · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 58.7 -> 57.9Hurt
Use beta=0.15 or 0.05 instead of 0.1 (ablation)Qwen3-4B-Base · Average over AIME24, AIME25, MATH500, OlympiadBenchaccuracy: 58.7 -> 57.6 (0.15); 58.0 (0.05)Hurt
Apply selection in science domain (OpenScienceReasoning_2)Qwen3-4B-Base · GPQA-Daccuracy: 39.9 -> 42.9Improved
Apply selection in science domainQwen3-4B-Base · MATH500accuracy: 85.8 -> 87.4Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy