Question Answering NLI on QNLI GLUE (test)
0.903AccuracyBERT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BERTBackbone=BERT-base-uncased2022.03 | 0.903 | 1 | |
| PoWER-BERTBackbone=BERT-base-uncased2022.03 | 0.897 | 1.23 | |
| AdapLeRBackbone=BERT-base-uncased2022.03 | 0.893 | 1.47 | |
| TR-BERTBackbone=BERT-base-uncased2022.03 | 0.89 | 1.09 | |
| DistilBERTBackbone=BERT-base-uncased2022.03 | 0.881 | 2 |