Topic Classification on Multi EURLEX (test)
94.8AccuracyBERT Seq-CLS
Evaluation results
| Method | Links | |
|---|---|---|
| BERT Seq-CLSTraining Strategy=individual2023.05 | 94.8 | |
| BERT Seq-CLSTraining Strategy=full2023.05 | 81.4 | |
| Binary BERTTraining Strategy=explicit2023.05 | 44.5 | |
| Bi-EncoderTraining Strategy=implicit2023.05 | 35.4 | |
| Bi-EncoderTraining Strategy=vanilla2023.05 | 33.5 | |
| Binary BERTTraining Strategy=implicit2023.05 | 30.3 | |
| Binary BERTTraining Strategy=vanilla2023.05 | 28.7 | |
| Bi-EncoderTraining Strategy=explicit2023.05 | 27.4 | |
| GPT-2Training Strategy=explicit2023.05 | 23.5 | |
| GPT-2Training Strategy=vanilla2023.05 | 22.3 | |
| BARTTraining Strategy=Zero-shot2023.05 | 19.8 | |
| GPT-2Training Strategy=implicit2023.05 | 18.6 |