Toxicity Mitigation on Offensive category
50.8RTROptimus
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| OptimusBackbone=LLaMA-2, Defense setting=Attack2025.07 | 50.8 | 5.31 | 9.8 | 60.6 | |
| OptimusBackbone=LLaMA-2, Defense setting=O-API, Healing type=Contextual (CH)2025.07 | 15.4 | 6.04 | 10.5 | 61.1 | |
| OptimusBackbone=LLaMA-2, Defense setting=No-attack2025.07 | 8.8 | 4.27 | 10.2 | 60.6 | |
| OptimusBackbone=LLaMA-2, Defense setting=P-API, Healing type=Contextual (CH)2025.07 | 6.2 | 5.84 | 10.8 | 61.1 | |
| OptimusBackbone=LLaMA-2, Defense setting=O-API, Healing type=Non-contextual (NH)2025.07 | 2.8 | 5.85 | 10.9 | 59.3 | |
| OptimusBackbone=LLaMA-2, Defense setting=Refusal, Healing type=Contextual (CH)2025.07 | 1.2 | 6.22 | 10.4 | 62.4 | |
| OptimusBackbone=LLaMA-2, Defense setting=P-API, Healing type=Non-contextual (NH)2025.07 | 0.8 | 5.89 | 11 | 59.2 | |
| OptimusBackbone=LLaMA-2, Defense setting=Refusal, Healing type=Non-contextual (NH)2025.07 | 0.2 | 5.96 | 10.5 | 59.8 |