ReAtt

Reported on 3 benchmarks across 2 tasks · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing3 results

Information RetrievalonNatural Questions
Precision@100· 2022-12-05
90.4
SOTA
Retrieval as Attention: End-to-end Learning of Retrieval and Reading within a Single Transformer arXiv:2212.02027
Information RetrievalonNatural Questions
Precision@20· 2022-12-05
86
SOTA
Retrieval as Attention: End-to-end Learning of Retrieval and Reading within a Single Transformer arXiv:2212.02027
Question AnsweringonNatural Questions
EM· 2022-12-05
54.7
best: 64 (Atlas (full, Wiki-dec-2018 index))
Retrieval as Attention: End-to-end Learning of Retrieval and Reading within a Single Transformer arXiv:2212.02027