Toxicity Classification on DC
31Harmonic Mean F1ToxiGAN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ToxiGANClassifier=RoBERTa2026.01 | 31 | 40.1 | |
| Ideal (Gold Label)Classifier=RoBERTa2026.01 | 30.7 | 39.7 | |
| ToxiGANClassifier=BERT2026.01 | 29.7 | 37.7 | |
| OversamplingClassifier=RoBERTa2026.01 | 29.5 | 38.6 | |
| Back-TranslateClassifier=RoBERTa2026.01 | 29.4 | 39.1 | |
| T5-ParaphaseClassifier=RoBERTa2026.01 | 29.4 | 38.9 | |
| EDAClassifier=RoBERTa2026.01 | 29.3 | 39 | |
| Base (No Aug.)Classifier=RoBERTa2026.01 | 29.2 | 38.6 | |
| Mistral-v0.3-SGClassifier=RoBERTa2026.01 | 29.1 | 38.3 | |
| GPT4o-TCClassifier=RoBERTa2026.01 | 28.7 | 38.1 | |
| Ideal (Gold Labels)Classifier=BERT2026.01 | 28.5 | 35.8 | |
| Mistral-v0.3-SGClassifier=BERT2026.01 | 28.5 | 37.6 | |
| Mistral-v0.3-ZGClassifier=BERT2026.01 | 28.2 | 37.4 | |
| Mistral-v0.3-ZGClassifier=RoBERTa2026.01 | 28.2 | 37.6 | |
| Mistral-v0.3-FSClassifier=RoBERTa2026.01 | 28.2 | 37.9 | |
| Mistral-v0.3-FSClassifier=BERT2026.01 | 28 | 37.6 | |
| OversamplingClassifier=BERT2026.01 | 27.8 | 35.3 | |
| GPT4o-TCClassifier=BERT2026.01 | 27.5 | 35.7 | |
| Llama3.2-TCClassifier=RoBERTa2026.01 | 27.5 | 37.3 | |
| Llama3.2-TCClassifier=BERT2026.01 | 27.4 | 37 | |
| GPT4.1-TCClassifier=RoBERTa2026.01 | 27.3 | 36.3 | |
| T5-ParaphaseClassifier=BERT2026.01 | 27.1 | 36.8 | |
| Back-TranslateClassifier=BERT2026.01 | 27 | 36.4 | |
| GPT4.1-TCClassifier=BERT2026.01 | 26.9 | 35.2 | |
| Base (No Aug.)Classifier=BERT2026.01 | 26.7 | 34.8 | |
| EDAClassifier=BERT2026.01 | 26.3 | 35.9 |