Toxicity Detection on TD Suite (Implicit/Adv/Toxigen)
60Implicit ScoreInfICL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| InfICLLLM=GPT3.5-turbo2026.04 | 60 | 64.13 | 84.5 | 69.54 | |
| DOPA*LLM=GPT4o-mini2026.04 | 59.5 | 65.25 | 83.25 | 69.33 | |
| InfICLLLM=GPT4o-mini2026.04 | 59.38 | 62.88 | 84.15 | 68.79 | |
| DOPA*LLM=GPT3.5-turbo2026.04 | 58.5 | 66.13 | 82.63 | 69.08 | |
| KNNLLM=GPT3.5-turbo2026.04 | 58.38 | 60.5 | 82.23 | 67.04 | |
| KNNLLM=GPT4o-mini2026.04 | 57.63 | 57.13 | 82 | 65.58 | |
| DICLLLM=GPT4o-mini2026.04 | 57.38 | 54.75 | 80.63 | 64.25 | |
| RewriteLLM=GPT4o-mini2026.04 | 54.63 | 54.75 | 72.25 | 60.54 | |
| RewriteLLM=GPT3.5-turbo2026.04 | 54.63 | 53.75 | 73.5 | 60.63 | |
| DICLLLM=GPT3.5-turbo2026.04 | 38.5 | 39.62 | 40.63 | 39.58 |