Toxicity Classification on Jigsaw dataset
44.2Rescue RateMistral-small-24B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Mistral-small-24BCondition=Aggregated across prompted conditions2026.05 | 44.2 | 43.1 | 8,536 | |
| DeepSeek-V3Condition=Aggregated across prompted conditions2026.05 | 38.1 | 36.9 | 7,264 | |
| Llama-3.1-8BCondition=Aggregated across prompted conditions2026.05 | 37.5 | 36.5 | 9,040 | |
| GPT-4o-miniCondition=Aggregated across prompted conditions2026.05 | 35.7 | 34.6 | 7,152 | |
| Mixtral-8x7BCondition=Aggregated across prompted conditions2026.05 | 34.6 | 33.5 | 7,416 | |
| Mistral-7BCondition=Aggregated across prompted conditions2026.05 | 31.6 | 30.5 | 7,568 | |
| Llama-3.1-70BCondition=Aggregated across prompted conditions2026.05 | 31.3 | 30.3 | 7,848 | |
| Llama-3.3-70BCondition=Aggregated across prompted conditions2026.05 | 30.4 | 29.4 | 7,592 | |
| Qwen-2.5-72bCondition=Aggregated across prompted conditions2026.05 | 27.8 | 26.8 | 6,504 |