Extractive Question Answering on Natural Questions MRQA
81F1 ScoreLinkBERT
Evaluation results
| Method | Links | |
|---|---|---|
| LinkBERTModel Scale=large2022.03 | 81 | |
| SplinterNumber of training examples=Full Dataset2021.01 | 81 | |
| SpanBERTNumber of training examples=Full Dataset2021.01 | 80.6 | |
| SpanBERT (Reimpl)Number of training examples=Full Dataset2021.01 | 80.5 | |
| RoBERTaNumber of training examples=Full Dataset2021.01 | 79.6 | |
| BERTModel Scale=large2022.03 | 79 | |
| LinkBERTModel Scale=base2022.03 | 78.3 | |
| BERTModel Scale=base2022.03 | 76.5 | |
| SplinterNumber of training examples=10242021.01 | 65.5 | |
| LinkBERTModel Scale=tiny2022.03 | 60.3 | |
| SpanBERT (Reimpl)Number of training examples=10242021.01 | 59.5 | |
| BERTModel Scale=tiny2022.03 | 58.9 | |
| SpanBERTNumber of training examples=10242021.01 | 57.5 | |
| RoBERTaNumber of training examples=10242021.01 | 54.2 | |
| SplinterNumber of training examples=1282021.01 | 46.3 | |
| SpanBERT (Reimpl)Number of training examples=1282021.01 | 36 | |
| SpanBERTNumber of training examples=1282021.01 | 32.2 | |
| RoBERTaNumber of training examples=1282021.01 | 30.1 | |
| SplinterNumber of training examples=162021.01 | 27.4 | |
| SpanBERTNumber of training examples=162021.01 | 19.7 | |
| SpanBERT (Reimpl)Number of training examples=162021.01 | 19.6 | |
| RoBERTaNumber of training examples=162021.01 | 17.3 |