Passage-Mask: A Learnable Regularization Strategy for Retriever-Reader Models

31060

Knowledge Triples

Source Entity	Relationship	Target Entity
Gradient Variance Control Method	uses	Learnable Passage Mask Method
Learnable Passage Mask Method	enhances	Retriever Reader Models Method
Learnable Passage Mask Method	outperforms	Baseline Method
Retriever Reader Models Method	appliesTo	Dialogue Conversation Task
Retriever Reader Models Method	appliesTo	Fact Verification Task
Retriever Reader Models Method	appliesTo	Open Question Answering Task