Toxicity Classification on Average across WZ, DC, HX, OR
48.8Harmonic F1Ideal (Gold Label)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ideal (Gold Label)Classifier=RoBERTa2026.01 | 48.8 | 57.6 | |
| ToxiGANClassifier=RoBERTa2026.01 | 48.4 | 57.3 | |
| Mistral-v0.3-FSClassifier=RoBERTa2026.01 | 47.9 | 56.5 | |
| Mistral-v0.3-SGClassifier=RoBERTa2026.01 | 47.8 | 56.2 | |
| T5-ParaphaseClassifier=RoBERTa2026.01 | 47.3 | 56.5 | |
| Back-TranslateClassifier=RoBERTa2026.01 | 47.2 | 56.3 | |
| EDAClassifier=RoBERTa2026.01 | 47 | 56.4 | |
| GPT4o-TCClassifier=RoBERTa2026.01 | 47 | 55.6 | |
| ToxiGANClassifier=BERT2026.01 | 46.7 | 55.2 | |
| OversamplingClassifier=RoBERTa2026.01 | 46.6 | 55.6 | |
| Mistral-v0.3-ZGClassifier=RoBERTa2026.01 | 46.6 | 55.5 | |
| Ideal (Gold Labels)Classifier=BERT2026.01 | 46.4 | 54.8 | |
| Base (No Aug.)Classifier=RoBERTa2026.01 | 46.4 | 55.2 | |
| Llama3.2-TCClassifier=RoBERTa2026.01 | 46.4 | 55.1 | |
| Back-TranslateClassifier=BERT2026.01 | 45.6 | 54.7 | |
| GPT4.1-TCClassifier=RoBERTa2026.01 | 45.6 | 54.3 | |
| Mistral-v0.3-FSClassifier=BERT2026.01 | 45.5 | 54.8 | |
| EDAClassifier=BERT2026.01 | 45.4 | 54.4 | |
| T5-ParaphaseClassifier=BERT2026.01 | 45.3 | 54 | |
| Llama3.2-TCClassifier=BERT2026.01 | 45.3 | 54.2 | |
| Mistral-v0.3-SGClassifier=BERT2026.01 | 44.8 | 54.3 | |
| OversamplingClassifier=BERT2026.01 | 44.6 | 53.6 | |
| Mistral-v0.3-ZGClassifier=BERT2026.01 | 44.4 | 54 | |
| Base (No Aug.)Classifier=BERT2026.01 | 44 | 53.2 | |
| GPT4o-TCClassifier=BERT2026.01 | 43.9 | 53.2 | |
| GPT4.1-TCClassifier=BERT2026.01 | 42.7 | 52.6 |