Research map / Label-efficient and robust learning / Knowledge distillation
Pedagogically-Inspired Data Synthesis for Language Model Knowledge Distillation
Bowei He, Yankai Chen, Xiaokun Zhang, Linghe Kong, Philip S. Yu, Xue Liu, Chen Ma
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Identifier (ablation) | IOA (Qwen2.5-3B) · DollyEval (o1 teacher) | ROUGE-L: 36.72 -> 38.16 | Improved |
| add Organizer curriculum (ablation) | IOA (Qwen2.5-3B) · GSM8K (o1 teacher) | Pass@1: 51.68 -> 55.79 | Improved |
| add Adapter (ablation) | IOA (Qwen2.5-3B) · HumanEval (o1 teacher) | Pass@1: 36.48 -> 40.64 | Improved |
| add Adapter (ablation) | IOA (LLaMA3.2-3B) · HumanEval (o1 teacher) | Pass@1: 34.42 -> 37.98 | Improved |
| add Organizer curriculum (ablation) | IOA (LLaMA3.2-3B) · GSM8K (o1 teacher) | Pass@1: 48.92 -> 52.07 | Improved |
| add Identifier (ablation) | IOA (LLaMA3.2-3B) · DollyEval (o1 teacher) | ROUGE-L: 35.47 -> 36.88 | Improved |
| add Identifier (ablation) | IOA (Qwen2.5-3B) · HumanEval (o1 teacher) | Pass@1: 39.87 -> 40.64 | Improved |
| use pedagogical IOA instead of best baseline MADA | Qwen2.5-3B · HumanEval (o1 teacher) | Pass@1: 33.39 -> 40.64 | Improved |
| switch teacher from o1 to DeepSeek-R1 | IOA (Qwen2.5-3B) · MATH | Pass@1: 15.53 -> 16.02 | Improved |
| use higher-quality 3K seed data | IOA (Qwen2.5-3B) · AIME2024 | Pass@1: 6.29 -> 27.52 | Improved |
| increase seed data amount to 9K synthesized samples | IOA (Qwen2.5-3B) · AIME2024 | Pass@1: 6.29 -> 11.83 | Improved |
| increase tau_ZPD beyond 0.15 | IOA · averaged ROUGE-L and Pass@1 | average | Mixed |
Closest papers by meaning
- MiniPLM: Knowledge Distillation for Pre-training Language Models (ICLR 2025 · similarity 0.92)
- DistiLLM: Towards Streamlined Distillation for Large Language Models (ICML 2024 · similarity 0.91)
- TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models (ICLR 2025 · similarity 0.91)
- DDK: Distilling Domain Knowledge for Efficient Large Language Models (NeurIPS 2024 · similarity 0.91)
- Distilling LLM Agent into Small Models with Retrieval and Code Tools (NeurIPS 2025 · similarity 0.90)
- Distilling What Matters: Confidence-Aware Selective Distillation for Large Language Models (NeurIPS 2026 · similarity 0.90)
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.
