Toxicity Classification
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
59.9Balanced Accuracy
57
Feb 26, 2026
48.8Harmonic F1
26
Feb 26, 2026
49.7Harmonic F1
26
Feb 26, 2026
44.2H.-F1
26
Feb 26, 2026
31Harmonic Mean F1
26
Feb 26, 2026
73.4Harmonic F1
26
Feb 26, 2026
94.35F1 Score
24
Feb 26, 2026
60.41Accuracy
22
Feb 26, 2026
99Class 0 Precision
19
Apr 21, 2026
91.47Accuracy
19
May 22, 2026
94.33Accuracy
19
May 22, 2026
72.3F1 Score
18
Jul 8, 2026
0.69Loss
16
Feb 26, 2026
0.61Loss
16
Feb 26, 2026
79.66Worst-Group Accuracy
13
Feb 26, 2026
75.3Worst-Group Accuracy
11
Apr 21, 2026
92.6Average Accuracy
10
Jun 19, 2026
44.2Rescue Rate
9
Jun 2, 2026
96Test Accuracy
9
Apr 8, 2026
86.9Hard Accuracy
7
Jun 19, 2026
98.53Accuracy
7
May 22, 2026
93.73Accuracy
7
May 22, 2026
91.33Accuracy
7
May 22, 2026
72.29Balanced ACC
6
May 18, 2026
67Macro F1
6
May 11, 2026