Intent Classification on NLU Eval (test)
88.9AccuracyBERT Seq-CLS
Evaluation results
| Method | Links | |
|---|---|---|
| BERT Seq-CLSTraining Strategy=individual2023.05 | 88.9 | |
| BERT Seq-CLSTraining Strategy=full2023.05 | 87.4 | |
| Binary BERTTraining Strategy=implicit2023.05 | 73.1 | |
| Binary BERTTraining Strategy=vanilla2023.05 | 71.6 | |
| Bi-EncoderTraining Strategy=vanilla2023.05 | 71.1 | |
| Bi-EncoderTraining Strategy=explicit2023.05 | 70 | |
| Binary BERTTraining Strategy=explicit2023.05 | 69.6 | |
| GPT-2Training Strategy=implicit2023.05 | 63.1 | |
| Bi-EncoderTraining Strategy=implicit2023.05 | 61.9 | |
| GPT-2Training Strategy=vanilla2023.05 | 61.7 | |
| GPT-2Training Strategy=explicit2023.05 | 60.7 | |
| BARTTraining Strategy=Zero-shot2023.05 | 40.1 | |
| GPT-3Training Strategy=Zero-shot2023.05 | 37.9 |