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

Safety Classification on HoliSafe-Bench

0.783AUROC

Qwen3-VL 8B

0.307720.431110.55450.67789May 1, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.05
0.783
2026.05
0.778
2026.05
0.768
2026.05
0.764
2026.05
0.762
2026.05
0.761
2026.05
0.758
2026.05
0.753
2026.05
0.75
2026.05
0.748
2026.05
0.744
2026.05
0.739
2026.05
0.737
2026.05
0.731
2026.05
0.728
2026.05
0.727
2026.05
0.725
2026.05
0.724
2026.05
0.72
2026.05
0.72
2026.05
0.718
2026.05
0.713
2026.05
0.701
2026.05
0.693
2026.05
0.689
2026.05
0.688
2026.05
0.688
2026.05
0.684
2026.05
0.668
2026.05
0.667
2026.05
0.663
2026.05
0.657
2026.05
0.65
2026.05
0.648
2026.05
0.638
2026.05
0.603
2026.05
0.599
2026.05
0.588
2026.05
0.584
2026.05
0.573
2026.05
0.511
2026.05
0.504
2026.05
0.499
2026.05
0.499
2026.05
0.49
2026.05
0.465
2026.05
0.39
2026.05
0.326
2026.05
0.326