Cross-task Generalization on Natural Instructions (test)
3.076Answerability ClassificationSkill-stratified
Evaluation Results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Skill-stratifiedSampling strategy=Skill-stratified, Evaluation setting=Out-of-domain2023.07 | 3.076 | 2.101 | 3.142 | 2.388 | 2.364 | 2.418 | 2.76 | 2.763 | 2.628 | 2.503 | 3.037 | 1.682 | 2.572 | |
| RandomSampling strategy=Random, Evaluation setting=Out-of-domain2023.07 | 3.048 | 2.068 | 3.101 | 2.363 | 2.329 | 2.399 | 2.744 | 2.749 | 2.591 | 2.472 | 3.027 | 1.665 | 2.546 | |
| SKILL-ITSampling strategy=SKILL-IT, Evaluation setting=Out-of-domain2023.07 | 3.043 | 2.067 | 3.099 | 2.359 | 2.32 | 2.389 | 2.733 | 2.733 | 2.586 | 2.468 | 3.015 | 1.668 | 2.54 |