AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Adversarial Data Augmentation for Task-Specific Knowledge Distillation of Pre-trained Transformers

Minjia Zhang, Niranjan Uma Naresh, Yuxiong He

AAAI 2022 · Architectures and efficiency › Pruning and efficient tuning · approach: Parameter-efficient transfer analysis · 11 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
Use adversarial embedding augmentation during task-specific distillationBERT12 → BERT6 · GLUE development setAvg: 82.4 -> 83.6Improved
Add task-specific AD2 distillationMiniLM-L12-H384 · GLUE development setAvg: 87.1 -> 87.6Improved
Add deep knowledge distillation (ablation)AD2, BERT12 → BERT6 · GLUE development setAvg: 85.9 -> 86.2Improved
Add knowledge distillation (ablation)AD2, BERT12 → BERT6 · GLUE development setAvg: 85.2 -> 85.9Improved
Add adversarial data augmentation (ablation)AD2, BERT12 → BERT6 · GLUE development setAvg: 84.3 -> 85.2Improved
Initialize the student from bottom rather than top layers (ablation)3-layer student initialized from DistilBERT · GLUE development setAvg: 75.6 -> 80.6Improved
Add AD2 to bottom-layer initialization (ablation)3-layer student initialized from DistilBERT · GLUE development setAvg: 80.6 -> 82.3Improved
Replace BERT12 with BERT24 as teacher while using KD without AD2 (ablation)BERT6 student · GLUE development setAvg: 84.2 -> 84.0No clear effect
Use BERT24 as teacher and add AD2 (ablation)BERT6 student · GLUE development setAvg: 84.0 -> 85.3Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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