LUKE

Reported on 5 benchmarks across 2 tasks · 2 papers · 1 SOTA

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

Natural Language Processing5 results

Relation ExtractiononTACRED
F1 (5% Few-Shot)· uses extra data· 2020-10-02
51.6
best: 69 (NLI_DeBERTa)
SOTA
LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention arXiv:2010.01057
Relation ExtractiononSemEval-2010 Task-8
F1· 2022-08-20
90.3
best: 91.9 (SP)
SPOT: Knowledge-Enhanced Language Representations for Information Extraction arXiv:2208.09625
Relation ExtractiononTACRED
F1 (1% Few-Shot)· uses extra data· 2020-10-02
17
best: 63.7 (NLI_DeBERTa)
LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention arXiv:2010.01057
Question AnsweringonSQuAD1.1 dev
EM· uses extra data· 2020-10-02
89.8
best: 90.06 (T5-11B)
LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention arXiv:2010.01057
Question AnsweringonSQuAD1.1
EM· 2020-10-02
90.2
best: 90.622 ({ANNA} (single model))
LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention arXiv:2010.01057