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

Toxicity Detection on Perturbed Text

85.57Performance (Insert)

ContiGuard

49.544458.897268.2577.6028Mar 16, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
85.5787.2385.1883.9790.482.9480.5892.1990.1586.473
77.264.6170.4860.5164.8463.2950.9366.3864.3764.7325.9
2026.03
69.5552.4758.7359.5852.6362.4450.2353.7166.9258.4828.3
66.1552.2461.1356.4950.7859.355184.4781.1462.5325.2
61.5963.7661.957.7356.9659.2756.0371.8775.6662.7529.8
2026.03
60.960.9758.8158.3453.5555.8758.3573.2675.1261.6914.4
60.258.556.4156.9682.5359.252.6366.6281.9263.8826.6
51.315153.3253.450.1551.9351.5578.5980.1457.9332.5
2026.03
51.3151.4751.8551.6249.9252.6351.1669.5572.5755.7935.8
2026.03
50.9350.3951.2452.3250.1550.3151.3955.2679.8354.6534