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

Backdoor Attack research benchmarks – Page 3 · SOTA2 Research

Backdoor Attack

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

GSM8K ⊕ BeaverTails

10ASR (No Trigger)

8

Jun 30, 2026

GSM8K ⊕ HEx-PHI

1.8ASR (Without Trigger)

8

Jun 30, 2026

Full fine-tuning

100ASR

8

Mar 5, 2026

ImageNet-1K (test)

94.4ASR

8

Mar 4, 2026

99ASR

6

Jun 23, 2026

96.2RTC

6

Mar 6, 2026

120.43LI

6

Feb 26, 2026

Legislative Drafting Scenario

2.23Average Turn

5

Jun 2, 2026

Healthcare Advice Scenario

3.14Average Turn Count

5

Jun 2, 2026

CodeAlpaca-20k (test)

90.5ASR

5

May 12, 2026

Wiki topic evaluation requests (test)

93.2ASR

5

May 12, 2026

shopping requests GPT-5-generated (test)

94.1Attack Success Rate (ASR)

5

May 12, 2026

CodeAlpaca 20k code generation instructions

94.8ASR

5

May 12, 2026

Wiki topic evaluation requests

94.5ASR

5

May 12, 2026

GPT-5-generated shopping requests

91.2ASR

5

May 12, 2026

2.07Mc Score

5

Feb 26, 2026

19.73Mc

5

Feb 26, 2026

23Mc Score

5

Feb 26, 2026

Backdoor Evaluation Set Out-of-merge

98.3Attack Success Rate

4

Jun 4, 2026

Backdoor Evaluation Set In-merge

100Attack Success Rate

4

Jun 4, 2026

Stable Diffusion 20,000 samples

VillanDiffusion

1,066Affected Parameters (M)

4

May 29, 2026

AdvBench (test)

91.3ASR

4

May 12, 2026

91.8ASR

4

May 12, 2026

Diffusion Models Image-to-image

Villan Diffusion

99.5ASR

4

May 8, 2026

77ASR

4

Mar 5, 2026

Page 3 of 4

Popular datasets

ImageNet CIFAR CelebA SD Fear Trigger Emotion

Follow for updates

Join SOTA2 Contact