MWE Paraphrasing on English MWE SemEval (test)
13.2P@1GPT-3
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-3# param=175B2023.06 | 13.2 | — | — | |
| T5-base# param=220M, rerank=false2023.06 | 10.8 | 22 | 25.8 | |
| T5-base# param=220M, rerank=true2023.06 | 10.1 | 22.6 | 27.1 | |
| SpanBERT# param=340M2023.06 | 8.5 | 22 | 28.9 | |
| BERT-base# param=110M2023.06 | 8.2 | 18.4 | 24.3 | |
| BERT-large# param=340M2023.06 | 8.1 | 19.4 | 27.6 | |
| T5-large# param=770M, rerank=false2023.06 | 8 | 22.5 | 28 | |
| T5-large# param=770M, rerank=true2023.06 | 7.2 | 25.6 | 33.3 | |
| ALBERT# param=17M2023.06 | 6.6 | 13.5 | 21.3 |