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

Multi-modal Toxicity Attack on PTP (PolygloToxicityPrompts) English (test)

5.1Overall Toxicity Rate

Text-Only

4.070811.017917.96524.9121May 1, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.05
5.14.82.75.81.5-2.6-4.4
2026.05
6.92.10.660.84.12--
2026.05
12.587.733.288.142.65-5.19-10.85
2026.05
15.838.474.7210.213.89-5.94-13.66
2026.05
17.37.61.6515.571.998.295.080.74-
2026.05
18.4210.565.3911.834.67-7.48-15.97
2026.05
19.669.112.57154.018.486.120.73-
2026.05
19.6811.295.8412.475.03-8.15-17.12
2026.05
20.4713.827.2912.356.14-7.51-19.63
2026.05
21.1313.858.4714.276.05-7.82-19.38
2026.05
21.711.632.8620.533.237.565.230.7-
2026.05
22.0111.061.2720.275.429.027.850.71-
2026.05
22.4513.638.8214.556.36-7.74-20.17
2026.05
22.6510.623.6421.13.837.696.920.71-
2026.05
23.2613.589.1515.347.67-7.89-21.83
2026.05
23.2714.1410.3815.598.28-8.28-21.65
2026.05
23.9214.9312.0515.868.94-8.61-22.19
2026.05
24.5812.052.6523.094.939.416.160.72-
2026.05
27.6812.384.1626.95.52107.210.65-
2026.05
27.9212.444.0224.855.919.727.140.73-
2026.05
28.1412.583.9625.386.799.487.080.74-
2026.05
30.1615.315.0528.257.4210.987.440.77-
2026.05
30.8314.564.1629.26.0610.227.060.74-