Toxicity Detection

Benchmarks

Dataset Name	SOTA Method	Metric
Toxicity Detection 64 model-persona combinations (8 models x 8 personas)	SVM Meta-Ensemble	Win Count58	56	1mo ago
ToxiGen	TaT (Disp.)	Score84.23	53	18d ago
Hosseini’s 1st Negated Sentence 1.0 (test)	Bert+Bi-LSTM	Toxicity Score80.6	24	1mo ago
HateXplain		AUC90.44	21	1mo ago
CivilComments-WILDS (test)		Average Accuracy92.7	19	1mo ago
Bangla		Accuracy91.66	15	5d ago
CivilComments (test)	JTT	WGA78.8	14	3d ago
Wikipedia Detox	GNN-based framework	Accuracy91.3	13	1mo ago
RealTox Challenging	Trajectory (Raw)	Accuracy98.92	12	1mo ago
RealTox Standard	Trajectory (Raw)	Accuracy87.56	12	1mo ago
CAD full	ContextRNN	F1 Score52.5	11	1mo ago
BAD		F1 Score80.8	11	1mo ago
BBF		F1 Score66.4	11	1mo ago
L2DTnH (test)	IGC-BERT	Accuracy96.05	10	4d ago
Perturbed Text	ContiGuard	Performance (Insert)85.57	10	1mo ago
Implicit Hate	EDA	Accuracy61.95	10	1mo ago
Hate Speech	I-Vanilla	Accuracy75.73	10	1mo ago
Civil	I-Simul	Accuracy87.06	10	1mo ago
CAD context	ContextRNN	F1 Score68.1	10	1mo ago
HQG	ContextSum	F1 Score91.1	10	1mo ago
HQR	ContextSingle	F1 Score81.1	10	1mo ago
FBK flipped	ContextSingle	F1 Score22	10	1mo ago
FBK full	ContextSingle	F1 Score41.3	10	1mo ago
ToxicChat	SafeGPT	F1 Score1	9	1mo ago
Bangla Toxicity Dataset		Accuracy Gap (AG)0	8	5d ago

Showing 25 of 52 rows