Question Answering on SQuADShifts 1.0 (test)
54.3Average F1multitask QAG
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| multitask QAGLanguage Model=BART-Large2023.05 | 54.3 | 36.7 | 53.6 | 34.3 | 54.1 | 36.4 | 57.7 | 41.8 | 51.6 | 34.4 | |
| end2end QAGLanguage Model=T5-Large2023.05 | 53.7 | 37.3 | 49.1 | 31.1 | 56.1 | 40.1 | 60.7 | 45.5 | 48.8 | 32.5 | |
| Gold QA2023.05 | 53.3 | 37.3 | 45.9 | 30.4 | 55.6 | 38.7 | 61.4 | 46.9 | 50.1 | 33.4 | |
| pipeline QAGLanguage Model=T5-Large2023.05 | 52 | 33.9 | 50.9 | 31.1 | 51.7 | 32.9 | 55.9 | 39.5 | 49.7 | 32 | |
| end2end QAGLanguage Model=T5-Base2023.05 | 51.8 | 35.4 | 44.9 | 26.9 | 56.9 | 40.1 | 59.9 | 45.3 | 45.3 | 29.2 | |
| pipeline QAGLanguage Model=BART-Large2023.05 | 51.7 | 34 | 49 | 30.2 | 52.5 | 33.7 | 55.3 | 40 | 49.7 | 32.3 | |
| pipeline QAGLanguage Model=T5-Base2023.05 | 51.7 | 33.6 | 50.6 | 31.1 | 52.8 | 34 | 53.8 | 37.1 | 49.6 | 32.4 | |
| multitask QAGLanguage Model=BART-Base2023.05 | 50.8 | 33.2 | 49.4 | 30.5 | 50.6 | 32.1 | 55 | 39.2 | 48.4 | 31.1 | |
| QG onlyLanguage Model=T5-Base2023.05 | 50.7 | 34.7 | 43.3 | 27.4 | 54.4 | 37.1 | 57.7 | 43.2 | 47.3 | 31.1 | |
| pipeline QAGLanguage Model=BART-Base2023.05 | 50 | 32.4 | 48.4 | 29.8 | 49.4 | 31.1 | 53 | 36 | 49.5 | 32.7 | |
| multitask QAGLanguage Model=T5-Base2023.05 | 49.6 | 30.9 | 48.8 | 28.9 | 48.1 | 28.5 | 52.5 | 35.1 | 49.1 | 31.2 | |
| multitask QAGLanguage Model=T5-Large2023.05 | 49.5 | 30.9 | 46.3 | 26 | 49.6 | 30.6 | 53 | 35.9 | 49 | 31.2 | |
| QG onlyLanguage Model=BART-Base2023.05 | 49.4 | 33.9 | 42.3 | 26.7 | 54.3 | 37.2 | 59.3 | 44.8 | 41.9 | 27 | |
| QG onlyLanguage Model=BART-Large2023.05 | 49.4 | 33.8 | 43.3 | 27.4 | 54 | 36.7 | 59.4 | 44.6 | 41.1 | 26.4 | |
| QG onlyLanguage Model=T5-Large2023.05 | 48.9 | 33.4 | 42.7 | 26.8 | 53.2 | 36.2 | 58.5 | 43.9 | 41.5 | 26.7 | |
| QG onlyLanguage Model=T5-Small2023.05 | 48.5 | 33.1 | 43.8 | 27.7 | 50.5 | 34.5 | 55.2 | 41 | 44.4 | 29.1 | |
| end2end QAGLanguage Model=T5-Small2023.05 | 48.3 | 31.7 | 42.3 | 24.8 | 54.7 | 37.2 | 55.2 | 40.1 | 41.1 | 24.8 | |
| pipeline QAGLanguage Model=T5-Small2023.05 | 45.4 | 27.4 | 41.6 | 23.8 | 46.9 | 27.2 | 48.9 | 32.1 | 44.1 | 26.6 | |
| multitask QAGLanguage Model=T5-Small2023.05 | 44 | 25.2 | 42.1 | 22.9 | 44.3 | 24.1 | 45.6 | 27.9 | 44 | 25.7 | |
| end2end QAGLanguage Model=BART-Base2023.05 | 34 | 21.4 | 29.3 | 16.5 | 35.4 | 23.2 | 44.6 | 31.1 | 26.6 | 15 | |
| end2end QAGLanguage Model=BART-Large2023.05 | 17.5 | 10.2 | 17.1 | 9.1 | 18.9 | 11.3 | 22.3 | 14.7 | 11.6 | 5.7 |