Machine Reading Comprehension on SQuAD (val)
89.4F1RoBERTa-Large + MUPPET
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RoBERTa-Large + MUPPETModel Architecture=RoBERTa-Large, Pre-finetuning=MUPPET, Task head reuse=true2021.01 | 89.4 | — | |
| RoBERTa-LargeModel Architecture=RoBERTa-Large, Task head reuse=true2021.01 | 88.7 | — | |
| ELECTRA-LargeModel Architecture=ELECTRA-Large2021.01 | 88.1 | — | |
| RoBERTa-Base + MUPPETModel Architecture=RoBERTa-Base, Pre-finetuning=MUPPET, Task head reuse=true2021.01 | 86.6 | — | |
| RoBERTa-BaseModel Architecture=RoBERTa-Base, Task head reuse=true2021.01 | 82.6 | — | |
| ELECTRA-BaseModel Architecture=ELECTRA-Base2021.01 | 80.5 | — |