Toxicity Analysis on RealToxicityPrompts Nontoxic
0.22Exp. Max. ToxicitySGEAT + DEXPERTS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SGEAT + DEXPERTSBackbone=1.3B LM, Category=Combined2022.02 | 0.22 | 0.07 | |
| DEXPERTSBackbone=1.3B LM, Category=Decoding-Time, Decoding Speed=3x slow2022.02 | 0.26 | 0.11 | |
| SGEAT + Rejection SamplingBackbone=1.3B LM, Category=Combined2022.02 | 0.26 | 0.11 | |
| SGEAT (augmented)Backbone=1.3B LM, Category=Domain-Adaptive Training, Prompt Design=augmented2022.02 | 0.37 | 0.26 | |
| SGEAT (standard)Backbone=1.3B LM, Category=Domain-Adaptive Training, Prompt Design=standard2022.02 | 0.38 | 0.28 | |
| Rejection SamplingBackbone=1.3B LM, Category=Decoding-Time, Decoding Speed=4x slow2022.02 | 0.38 | 0.28 | |
| SGEAT (heuristic)Backbone=1.3B LM, Category=Domain-Adaptive Training, Prompt Design=heuristic2022.02 | 0.4 | 0.31 | |
| DAPT (nontoxic)Backbone=1.3B LM, Category=Domain-Adaptive Training2022.02 | 0.41 | 0.33 | |
| Word BanningBackbone=1.3B LM, Category=Decoding-Time2022.02 | 0.49 | 0.47 | |
| Jigsaw (nontoxic)Backbone=1.3B LM, Category=Domain-Adaptive Training2022.02 | 0.53 | 0.53 |