List your company
ResearchFormerly WizWand

Hyper-relational extraction on HyperRED (dev)

91.9PrecisionText2NKG
8.86463230.42189151.9791573.536409Nov 18, 2022Jan 11, 2023Mar 6, 2023Apr 29, 2023Jun 22, 2023Aug 15, 2023Oct 8, 2023
Updated 8mo ago

Evaluation results

MethodLinks
Text2NKGPLM=BERT-large (340M), Training Paradigm=Supervised2023.1091.979.4385.21
Text2NKGPLM=BERT-base (110M), Training Paradigm=Supervised2023.1091.2679.3684.89
Text2NKG w/o alphaPLM=BERT-base (110M), Training Paradigm=Supervised2023.1088.7778.3983.26
Text2NKG w/o DAPLM=BERT-base (110M), Training Paradigm=Supervised2023.1076.0272.2874.1
Pipeline Baseline (Large)Parameters=680M2022.1170.5866.5868.52
Pipeline BaselinePLM=BERT-large (340M), Training Paradigm=Supervised2023.1070.5866.5868.52
Pipeline Baseline (Medium)Parameters=221M2022.1169.762.3365.8
Pipeline BaselineModel Scale=Base, Parameters=132M2022.1169.2358.2163.24
Pipeline BaselineParameters=132M2022.1169.2358.2163.24
Pipeline BaselinePLM=BERT-base (110M), Training Paradigm=Supervised2023.1069.2358.2163.24
CubeREModel Scale=Large, Parameters=343M2022.1168.7568.8868.81
CubeRE (Large)Parameters=343M2022.1168.7568.8868.81
CubeREPLM=BERT-large (340M), Training Paradigm=Supervised2023.1068.7568.8868.81
Generative BaselineModel Scale=Large, Parameters=400M2022.1167.0865.7366.4
Generative Baseline (Large)Parameters=400M2022.1167.0865.7366.4
Generative BaselinePLM=BERT-large (340M), Training Paradigm=Supervised2023.1067.0865.7366.4
CubeREModel Scale=Base, Parameters=115M2022.1166.1464.3965.24
CubeREParameters=115M2022.1166.1464.3965.24
CubeREPLM=BERT-base (110M), Training Paradigm=Supervised2023.1066.1464.3965.23
Generative BaselineModel Scale=Base, Parameters=140M2022.1163.7959.9461.8
Generative BaselineParameters=140M2022.1163.7959.9461.8
Generative BaselinePLM=BERT-base (110M), Training Paradigm=Supervised2023.1063.7959.9461.8
Text2NKG w/o HMPLM=BERT-base (110M), Training Paradigm=Supervised2023.1061.7476.9768.52
GPT-4PLM=gpt-4, Training Paradigm=Unsupervised2023.1015.732415.237715.4811
ChatGPTPLM=gpt-3.5-turbo, Training Paradigm=Unsupervised2023.1012.058311.276411.6542