| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| RealToxicityPrompts | Toxicity Score0 | 29 | 3mo ago | ||
| BOLD | Toxic Rate0 | 26 | 2d ago | ||
| BOLD 23679 prompts (test) | DeStein | Avg Toxicity (Max)0.02 | 18 | 3mo ago | |
| AttaQ | RAD | Max Toxicity Score0.04 | 14 | 3mo ago | |
| AttaQ 1402 prompts (test) | RAD | Max Toxicity Score0.042 | 14 | 3mo ago | |
| RealToxicityPrompts 1K non-toxic prompts, 1K toxic prompts | Count of Non-Toxic Samples5 | 14 | 3mo ago | ||
| RealToxicity | Model Surgery | RealTox5.17 | 8 | 3mo ago | |
| Counterfactual Open-Ended (OCF) | Toxic Fraction0 | 5 | 1mo ago | ||
| DialogSum (DS) | Toxic Fraction0 | 5 | 1mo ago | ||
| RealToxicityPrompts RTP-N (Nontoxic) | Toxic Fraction0.2 | 5 | 1mo ago | ||
| RealToxicityPrompts RTP-C | Toxic Fraction18.1 | 5 | 1mo ago | ||
| RealToxicityPrompts responses | CodeGen-Multi-16B | Classifier Score21 | 3 | 3mo ago |
