RoBERTa-Base Joint MSPP

Reported on 5 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 Processing5 results

Question AnsweringonASNQ
MAP· 2022-05-02
0.673
best: 0.743 (DeBERTa-V3-Large + SSP)
SOTA
Paragraph-based Transformer Pre-training for Multi-Sentence Inference arXiv:2205.01228
Question AnsweringonASNQ
MRR· 2022-05-02
0.737
best: 0.8 (DeBERTa-V3-Large + SSP)
SOTA
Paragraph-based Transformer Pre-training for Multi-Sentence Inference arXiv:2205.01228
Question AnsweringonWikiQA
MAP· 2022-05-02
0.887
best: 0.927 (TANDA-DeBERTa-V3-Large + ALL)
Paragraph-based Transformer Pre-training for Multi-Sentence Inference arXiv:2205.01228
Question AnsweringonWikiQA
MRR· 2022-05-02
0.9
best: 0.939 (TANDA-DeBERTa-V3-Large + ALL)
Paragraph-based Transformer Pre-training for Multi-Sentence Inference arXiv:2205.01228
Fact VerificationonFEVER
Accuracy· 2022-05-02
74.39
best: 79.47 (ProoFVer-SB)
Paragraph-based Transformer Pre-training for Multi-Sentence Inference arXiv:2205.01228