Toxicity Classification on OR
49.7Harmonic F1Ideal (Gold Labels)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ideal (Gold Labels)Classifier=BERT2026.01 | 49.7 | 63.8 | |
| Ideal (Gold Label)Classifier=RoBERTa2026.01 | 49.4 | 63.7 | |
| ToxiGANClassifier=RoBERTa2026.01 | 48.3 | 62.9 | |
| ToxiGANClassifier=BERT2026.01 | 47.8 | 62.7 | |
| T5-ParaphaseClassifier=BERT2026.01 | 47.6 | 62.2 | |
| Back-TranslateClassifier=BERT2026.01 | 47.5 | 62.2 | |
| Mistral-v0.3-FSClassifier=RoBERTa2026.01 | 47.5 | 62.2 | |
| OversamplingClassifier=BERT2026.01 | 47.2 | 62 | |
| EDAClassifier=BERT2026.01 | 47.2 | 62 | |
| Llama3.2-TCClassifier=BERT2026.01 | 47.1 | 61.6 | |
| T5-ParaphaseClassifier=RoBERTa2026.01 | 46.9 | 61.8 | |
| EDAClassifier=RoBERTa2026.01 | 46.8 | 62 | |
| Mistral-v0.3-FSClassifier=BERT2026.01 | 46.6 | 61.5 | |
| GPT4o-TCClassifier=RoBERTa2026.01 | 46.6 | 61.5 | |
| Base (No Aug.)Classifier=BERT2026.01 | 46.5 | 61.6 | |
| GPT4.1-TCClassifier=BERT2026.01 | 46.2 | 60.9 | |
| GPT4o-TCClassifier=BERT2026.01 | 46.2 | 61.1 | |
| OversamplingClassifier=RoBERTa2026.01 | 46.2 | 61.4 | |
| Mistral-v0.3-SGClassifier=RoBERTa2026.01 | 46.1 | 61.5 | |
| Back-TranslateClassifier=RoBERTa2026.01 | 45.9 | 61.1 | |
| Base (No Aug.)Classifier=RoBERTa2026.01 | 45.6 | 61 | |
| Mistral-v0.3-SGClassifier=BERT2026.01 | 45.5 | 61 | |
| GPT4.1-TCClassifier=RoBERTa2026.01 | 45.5 | 60.6 | |
| Mistral-v0.3-ZGClassifier=RoBERTa2026.01 | 44.9 | 60.6 | |
| Mistral-v0.3-ZGClassifier=BERT2026.01 | 44.6 | 60.2 | |
| Llama3.2-TCClassifier=RoBERTa2026.01 | 44.5 | 59.7 |