RealToxicityPrompts
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
RealToxicityPrompts RTP-C
18.1Toxic Fraction
5
RealToxicityPrompts RTP-N
0.006Sentiment Parity
5
RealToxicityPrompts RTP-C
0.2Sentiment Parity
5
RealToxicityPrompts Non-toxic
2.8Stereotype Fraction
5
RealToxicityPrompts RTP-C Toxic
8.2Stereotype Fraction
5
REALTOXICITYPROMPTS (test)
0.081Toxicity Score (Avg)
5
RealToxicityPrompts 1.0 (Toxic)
0.299Toxicity
5
RealToxicityPrompts 1.0 (Random)
0.122Toxicity
5
RealToxicityPrompts benign
0.555BERTScore
4
RealToxicityPrompts Non-toxic prompts (test)
7.38Toxicity Probability
4
RealToxicityPrompts
0.265Avg Max Toxicity Score
4
RealToxicityPrompts 100k prompts
10.4Toxicity Score (Basic)
4
RealToxicityPrompts toxic-injection study data audit (train)
22.73Estimated Toxic Percentage
3
RealToxicityPrompts (test)
96Safety Score
3
RealToxicityPrompts responses
21Classifier Score
3
RealToxicityPrompts Toxic prompts (test)
74.29Toxicity Probability
2
REALTOXICITYPROMPTS
10Inappropriate Probability
2