Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Adversarial Robustness research benchmarks – Page 2 · SOTA2 Research

Adversarial Robustness

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

56.25DR

12

Feb 26, 2026

MUSE-Book Harry Potter

1.5ASR

11

Apr 20, 2026

Adversarial Robustness Prompts

92GCG ASR

11

Feb 26, 2026

4.69ASR

11

Feb 26, 2026

CIFAR-10 L2, epsilon=8/255

Stutz et al. (2020)

77.64Robust Acc (AA)

11

Feb 26, 2026

CIFAR-10 l_inf, epsilon=8/255 (test)

92.44Clean Accuracy

10

Jun 18, 2026

Base (8B Instruct)

17DirReq

10

May 12, 2026

RoboPAIR Adversarial Behaviors

92.3Attack Success Rate

10

Mar 5, 2026

82.07Baseline Acc

10

Feb 26, 2026

CIFAR-10 L-infinity, epsilon=4/255

Xiao et al. (2020) ADV

64.91Robust Accuracy (AA)

10

Feb 26, 2026

CIFAR-100 adv (test)

22.78Adversarial Error Rate

9

Jun 16, 2026

Defended (ours)

6.2ASR

9

May 25, 2026

Jailbreak Structured-based

2.62Perplexity

9

Mar 24, 2026

Jailbreak Perturbation-based

2.58Perplexity

9

Mar 24, 2026

Toxicity Perturbation-based

9.52Perplexity

9

Mar 24, 2026

3.12ASR

9

Feb 26, 2026

50.16ASR

9

Feb 26, 2026

CIFAR-100 sr=90% (test)

64.37Clean Accuracy

9

Feb 26, 2026

CIFAR-100 l_inf, epsilon=8/255 (test)

73.85Clean Accuracy

9

Feb 26, 2026

85.92Clean Accuracy

8

Jun 4, 2026

59.1Risk Ratio

8

May 12, 2026

28.3Risk Ratio

8

May 12, 2026

69.2Risk Ratio

8

May 12, 2026

40.9Risk Ratio

8

May 12, 2026

6ASR

8

May 8, 2026

Page 2 of 4

Popular datasets

CIFAR ImageNet CIFAR-100 CIFAR-10 Surface-form evasion probe suite

Follow for updates

Join SOTA2 Contact