Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Large Audio-Language Model Safety Evaluation on Figstep audio

62.8ASR

MDSteer-h2s

-0.01616.29232.648.908Oct 20, 2025
Updated 26d ago

Evaluation Results

MethodLinks
2025.10
62.865.60
2025.10
58.850.80
2025.10
52.353.80.8
2025.10
50.865.40
2025.10
4271.80
2025.10
35.265.812
2025.10
3471.40.4
2025.10
31.280.20.8
2025.10
3069.20.8
2025.10
28.273.510.8
2025.10
26.867.60.8
2025.10
25.268.654.4
2025.10
15.273.42
2025.10
7.295.31
2025.10
6.489.812.4
2025.10
5.2941.6
2025.10
3.670.655.2
2025.10
2.466.454