Retrieval as Attention: End-to-end Learning of Retrieval and Reading within a Single Transformer
Open paper
Benchmark
Model
Rank
Results
passage-retrieval-on-natural-questions
ReAtt
#1
Precision@100: 90.40
Precision@20: 86.00
question-answering-on-natural-questions
ReAtt
#6
EM: 54.7
Rank counts only results with a code link.