AdvGLUE

Adversarial GLUE

Dataset Information
Modalities
Texts
Languages
English
Introduced
2021
License
CC BY-SA 4.0
Homepage
Official Website
Contents

Overview

Adversarial GLUE (AdvGLUE) is a new multi-task benchmark to quantitatively and thoroughly explore and evaluate the vulnerabilities of modern large-scale language models under various types of adversarial attacks. In particular, we systematically apply 14 textual adversarial attack methods to GLUE tasks to construct AdvGLUE, which is further validated by humans for reliable annotations.

Description from: Adversarial GLUE: A Multi-Task Benchmark for Robustness Evaluation of Language Models

Variants: AdvGLUE

Associated Benchmarks

This dataset is used in 1 benchmark:

Recent Benchmark Submissions

Task Model Paper Date
Adversarial Robustness DeBERTa (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness ALBERT (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness T5 (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness SMART_RoBERTa (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness FreeLB (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness RoBERTa (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness InfoBERT (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness ELECTRA (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness BERT (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04
Adversarial Robustness SMART_BERT (single model) Adversarial GLUE: A Multi-Task Benchmark … 2021-11-04

Research Papers

Recent papers with results on this dataset:

External Links: