Text Classification on UTCD in-domain 1.0 (test)
63Go EmotionsBERT Seq-CLS
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| BERT Seq-CLSTraining Strategy=individual2023.05 | 63 | 69.5 | 92.9 | 78.7 | 95.2 | 85.5 | 94.1 | 99.2 | 73.4 | 83.4 | |
| Binary BERTTraining Strategy=explicit (ours)2023.05 | 60.2 | 66.6 | 91.8 | 93.4 | 88 | 80.4 | 88.7 | 98.9 | 67.8 | 81.7 | |
| GPT-2Training Strategy=explicit (ours)2023.05 | 60.1 | 70.1 | 76.4 | 84.3 | 81.9 | 76.7 | 87.9 | 98.6 | 68.6 | 78.3 | |
| Binary BERTTraining Strategy=implicit (ours)2023.05 | 59.9 | 67.2 | 91.8 | 93.5 | 87.1 | 81.8 | 89.2 | 98.9 | 68.1 | 82 | |
| Binary BERTTraining Strategy=vanilla2023.05 | 59.3 | 67.6 | 92.4 | 91.5 | 87.8 | 81.8 | 90 | 98.9 | 67.9 | 81.9 | |
| Bi-EncoderTraining Strategy=vanilla2023.05 | 59.2 | 65.7 | 92.8 | 82.2 | 84.4 | 79.9 | 89.3 | 99 | 67.4 | 80 | |
| GPT-2Training Strategy=implicit (ours)2023.05 | 59 | 70.3 | 71.4 | 84.7 | 81.7 | 73.1 | 87.7 | 98.4 | 68.3 | 77.2 | |
| Bi-EncoderTraining Strategy=explicit (ours)2023.05 | 58.8 | 66.8 | 91.8 | 82.7 | 83.3 | 79.9 | 89.5 | 98.9 | 67.7 | 80 | |
| GPT-2Training Strategy=vanilla2023.05 | 58.8 | 70.6 | 75.9 | 84.2 | 81.4 | 75.3 | 86.7 | 98.5 | 68.3 | 77.7 | |
| Bi-EncoderTraining Strategy=implicit (ours)2023.05 | 56.9 | 66 | 90.9 | 81.3 | 82 | 78.9 | 88.8 | 99 | 67.9 | 79.1 | |
| BERT Seq-CLSTraining Strategy=full2023.05 | 56.7 | 55.4 | 91.1 | 80.5 | 82.9 | 77.3 | 86.7 | 98.6 | 66.6 | 77.3 | |
| BARTTraining Strategy=Zero-shot2023.05 | 24.2 | 47.8 | 37.7 | 41.4 | 50.4 | 27.5 | 71.7 | 65 | 49.2 | 46.1 |