ClarifyDelphi: Reinforced Clarification Questions with Defeasibility Rewards for Social and Moral Situations
38037Knowledge Triples
| Source Entity | Relationship | Target Entity |
|---|---|---|
|
ClarifyDelphi
OtherEntity |
uses |
Обучение с подкреплением
Method |
|
Когнитивная наука
OtherEntity |
informs |
Моральная оценка
Task |
|
Обучение с подкреплением
Method |
employs |
Награда за отменяемость
Metric |