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

Toxicity Defense on MiniGPT-4

2.89Toxicity Score

NullSteer

0.817614.806328.79542.7837Mar 23, 2026
Updated 2mo ago

Evaluation Results

MethodLinks
2026.03
2.89
2026.03
3.26
2026.03
4.48
2026.03
4.51
2026.03
6.45
2026.03
8.84
2026.03
9.52
2026.03
11.3
2026.03
16.51
2026.03
18.93
2026.03
19.73
2026.03
20.93
2026.03
21.23
2026.03
22.78
2026.03
25.03
2026.03
25.76
2026.03
30.1
2026.03
30.28
2026.03
30.65
2026.03
30.65
2026.03
30.65
2026.03
30.65
2026.03
31.99
2026.03
32.42
2026.03
34.59
2026.03
35.71
2026.03
38.54
2026.03
38.97
2026.03
39.73
2026.03
42.86
2026.03
45.15
2026.03
48.52
2026.03
48.71
2026.03
50.12
2026.03
52.12
2026.03
54.7