KELM (finetuning BERT-large based single model)

Reported on 5 benchmarks across 2 tasks · 1 paper

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

Natural Language Processing5 results

Question AnsweringonCOPA
Accuracy· 2021-09-09
78
best: 100 (PaLM 540B (finetuned) )
KELM: Knowledge Enhanced Pre-Trained Language Representations with Message Passing on Hierarchical Relational Graphs arXiv:2109.04223
Question AnsweringonMultiRC
EM· 2021-09-09
27.2
best: 69.2 (PaLM 540B (finetuned) )
KELM: Knowledge Enhanced Pre-Trained Language Representations with Message Passing on Hierarchical Relational Graphs arXiv:2109.04223
Question AnsweringonMultiRC
F1· 2021-09-09
70.8
best: 90.1 (PaLM 540B (finetuned) )
KELM: Knowledge Enhanced Pre-Trained Language Representations with Message Passing on Hierarchical Relational Graphs arXiv:2109.04223
Common Sense ReasoningonReCoRD
EM· 2021-09-09
76.2
best: 95.9 (Turing NLR v5 XXL 5.4B (fine-tuned))
KELM: Knowledge Enhanced Pre-Trained Language Representations with Message Passing on Hierarchical Relational Graphs arXiv:2109.04223
Common Sense ReasoningonReCoRD
F1· 2021-09-09
76.7
best: 96.4 (Turing NLR v5 XXL 5.4B (fine-tuned))
KELM: Knowledge Enhanced Pre-Trained Language Representations with Message Passing on Hierarchical Relational Graphs arXiv:2109.04223