Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Adversarial Robustness on Toxicity Perturbation-based
Loading...
9.52
Perplexity
NullSteer
4.2832
39.6316
74.98
110.3284
Mar 23, 2026
Perplexity
Updated 25d ago
Evaluation Results
Method
Method
Links
Perplexity
NullSteer
Base Model=MiniGPT-4
2026.03
9.52
ASTRA
Base Model=MiniGPT-4
2026.03
10.14
NullSteer
Base Model=Qwen2-VL
2026.03
38.45
ASTRA
Base Model=Qwen2-VL
2026.03
40.14
Vanilla
Base Model=MiniGPT-4
2026.03
51.42
NullSteer
Base Model=LLaVA-v1.5
2026.03
57.96
ASTRA
Base Model=LLaVA-v1.5
2026.03
59.28
Vanilla
Base Model=LLaVA-v1.5
2026.03
63.68
Vanilla
Base Model=Qwen2-VL
2026.03
140.44
Feedback
Search any
task
Search any
task