Research map / Label-efficient and robust learning / Knowledge distillation
Learning to Augment for Data-scarce Domain BERT Knowledge Distillation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use L2A augmentation and student-feedback selection instead of the strongest listed baseline | BERT_Tiny distillation · SciTail | accuracy: 0.7512 -> 0.7827 | Helped |
| Use L2A augmentation and student-feedback selection instead of the strongest listed baseline | BERT_Tiny distillation · CIKM PI | accuracy: 0.9060 -> 0.9195 | Helped |
| Use L2A augmentation and student-feedback selection instead of the strongest listed baseline | BERT_Tiny distillation · RT | accuracy: 0.7366 -> 0.7798 | Helped |
| Use L2A augmentation and student-feedback selection instead of the strongest listed baseline | BERT_Tiny distillation · Amazon Watches | Pearson correlation: 0.4680 -> 0.4852 | Helped |
| Add target-domain data to augmentation (ablation) | L2A · SciTail | accuracy: 0.7714 -> 0.7827 | Helped |
| Add target-domain data to augmentation (ablation) | L2A · CIKM PI | accuracy: 0.8549 -> 0.9195 | Helped |
| Add source-domain data to augmentation (ablation) | L2A · RT | accuracy: 0.7526 -> 0.7798 | Helped |
| Add source-domain data to augmentation (ablation) | L2A · Amazon Watches | Pearson correlation: 0.4757 -> 0.4852 | Helped |
| Add attention-based distillation loss (ablation) | L2A · RT | accuracy: 0.7563 -> 0.7798 | Helped |
| Add hidden-representation distillation loss (ablation) | L2A · RT | accuracy: 0.7629 -> 0.7798 | Helped |
| Add dark-knowledge distillation loss (ablation) | L2A · RT | accuracy: 0.7647 -> 0.7798 | Helped |
| Use Electronics rather than Home as the source domain (ablation) | L2A · Amazon Watches | performance: +0.5% | Helped |
Closest papers by meaning
- DistiLLM: Towards Streamlined Distillation for Large Language Models (ICML 2024 · similarity 0.87)
- TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models (ICLR 2025 · similarity 0.86)
- DA-KD: Difficulty-Aware Knowledge Distillation for Efficient Large Language Models (ICML 2025 · similarity 0.86)
- MiniPLM: Knowledge Distillation for Pre-training Language Models (ICLR 2025 · similarity 0.86)
- DDK: Distilling Domain Knowledge for Efficient Large Language Models (NeurIPS 2024 · similarity 0.86)
- How to Trade Off the Quantity and Capacity of Teacher Ensemble: Learning Categorical Distribution to Stochastically Employ a Teacher for Distillation (AAAI 2024 · similarity 0.86)
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.