Natural Language Inference on ANLI MNLI SNLI FEVER (dev)
0.738R1 ScoreRoBERTa
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RoBERTaSize=LARGE2020.04 | 0.738 | 0.489 | 0.444 | 0.537 | |
| ALUMROBERTASize=LARGE2020.04 | 0.733 | 0.534 | 0.482 | 0.577 | |
| XLNetSize=LARGE2020.04 | 0.676 | 0.507 | 0.483 | 0.551 | |
| ALUMBERTSize=BASE2020.04 | 0.62 | 0.486 | 0.481 | 0.526 | |
| BERT+Size=BASE2020.04 | 0.575 | 0.473 | 0.43 | 0.489 | |
| BERTSize=LARGE2020.04 | 0.574 | 0.483 | 0.435 | 0.493 | |
| BERTSize=BASE2020.04 | 0.557 | 0.463 | 0.434 | 0.482 |