Relation Extraction on TACRED
86.6Micro F1RAG4RE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RAG4REbackbone=Flan T5-XL2024.04 | 86.6 | — | — | |
| DeepStruct2024.04 | 76.8 | — | — | |
| DEEPEXZero-Shot=true, Top-10=true2021.09 | 76.4 | — | — | |
| EXOBRAIN2024.04 | 75 | — | — | |
| SP2024.04 | 74.8 | — | — | |
| GAP2024.04 | 72.7 | — | — | |
| K-ADAPTERadapters=facAdapter + linAdapter2020.02 | 72.04 | 70.14 | 74.04 | |
| K-ADAPTERadapters=linAdapter2020.02 | 71.96 | 68.85 | 75.37 | |
| TANL*Supervised=true2021.09 | 71.9 | — | — | |
| TANLTraining Setting=standard2021.01 | 71.9 | — | — | |
| K-ADAPTERadapters=facAdapter2020.02 | 71.89 | 69.39 | 74.59 | |
| KEPLER2020.02 | 71.7 | 70.43 | 73.02 | |
| RoBERTamode=multitask2020.02 | 71.62 | 70.18 | 73.11 | |
| BERT+MTB2020.02 | 71.5 | — | — | |
| KnowBERT2020.02 | 71.5 | 71.6 | 71.4 | |
| BERTEM+MTB*Supervised=true2021.09 | 71.5 | — | — | |
| DG-SpanBERT*Supervised=true2021.09 | 71.5 | — | — | |
| BERT-EM+MTBTraining Setting=standard2021.01 | 71.5 | — | — | |
| DG-SpanBERTTraining Setting=standard2021.01 | 71.5 | — | — | |
| KnowBERT2020.12 | 71.5 | 71.6 | 71.4 | |
| RoBERTa2020.02 | 71.25 | 70.17 | 72.36 | |
| SpanBERT2020.12 | 70.8 | 70.8 | 70.9 | |
| GDPNet2020.12 | 70.5 | 72 | 69 | |
| BERT-large2020.02 | 70.1 | — | — | |
| BERT-EM*Supervised=true2021.09 | 70.1 | — | — | |
| BERT-EMTraining Setting=standard2021.01 | 70.1 | — | — | |
| CPSupervised training data percentage=100%, Input configuration=C+M2020.10 | 69.5 | — | — | |
| MTBSupervised training data percentage=100%, Input configuration=C+M2020.10 | 69.1 | — | — | |
| TANLTraining Setting=multi-task2021.01 | 69.1 | — | — | |
| LST-AGCN2020.12 | 68.8 | — | — | |
| BERTSupervised training data percentage=100%, Input configuration=C+M2020.10 | 68.3 | — | — | |
| C-AGGCN2020.12 | 68.2 | 73.1 | 60.9 | |
| ERNIE2020.02 | 67.97 | 69.97 | 66.08 | |
| C-GCN2020.02 | 66.4 | 69.9 | 63.3 | |
| BERT-base2020.02 | 66 | 67.23 | 64.81 | |
| PA-LSTM2020.12 | 65.1 | 65.7 | 64.5 | |
| CPSupervised training data percentage=10%, Input configuration=C+M2020.10 | 63.3 | — | — | |
| LSTM2020.12 | 62.7 | 65.7 | 59.9 | |
| MTBSupervised training data percentage=10%, Input configuration=C+M2020.10 | 60.8 | — | — | |
| CPSupervised training data percentage=100%, Input configuration=OnlyC2020.10 | 59.3 | — | — | |
| MTBSupervised training data percentage=100%, Input configuration=OnlyC2020.10 | 58.1 | — | — | |
| BERTSupervised training data percentage=10%, Input configuration=C+M2020.10 | 57.9 | — | — | |
| BERTSupervised training data percentage=100%, Input configuration=OnlyC2020.10 | 57 | — | — | |
| CPSupervised training data percentage=10%, Input configuration=OnlyC2020.10 | 51.5 | — | — | |
| MTBSupervised training data percentage=10%, Input configuration=OnlyC2020.10 | 49.6 | — | — | |
| DEEPEXZero-Shot=true, Top-1=true2021.09 | 49.2 | — | — | |
| CPSupervised training data percentage=1%, Input configuration=C+M2020.10 | 48.5 | — | — | |
| BERTSupervised training data percentage=100%, Input configuration=OnlyM2020.10 | 46.6 | — | — | |
| CPSupervised training data percentage=10%, Input configuration=OnlyM2020.10 | 45.3 | — | — | |
| CPSupervised training data percentage=100%, Input configuration=OnlyM2020.10 | 45 | — | — | |
| BERTSupervised training data percentage=10%, Input configuration=OnlyC2020.10 | 44.6 | — | — | |
| MTBSupervised training data percentage=10%, Input configuration=OnlyM2020.10 | 44.1 | — | — | |
| BERTSupervised training data percentage=10%, Input configuration=OnlyM2020.10 | 43.3 | — | — | |
| MTBSupervised training data percentage=100%, Input configuration=OnlyM2020.10 | 43.3 | — | — | |
| CPSupervised training data percentage=1%, Input configuration=OnlyC2020.10 | 39.3 | — | — | |
| Ours (mix-OBO)K=32, Base Model=RetrievalRE2025.05 | 37.46 | — | — | |
| Ours (mix-AAO)K=32, Base Model=RetrievalRE2025.05 | 37.23 | — | — | |
| RetrievalREK=322025.05 | 37.07 | — | — | |
| Ours (mix-AAO)K=16, Base Model=RetrievalRE2025.05 | 36.57 | — | — | |
| KnowPromptK=322025.05 | 36.33 | — | — | |
| Ours (mix-OBO)K=32, Base Model=KnowPrompt2025.05 | 36.25 | — | — | |
| Ours (mix-OBO)K=16, Base Model=RetrievalRE2025.05 | 35.83 | — | — | |
| Ours (mix-AAO)K=32, Base Model=KnowPrompt2025.05 | 35.66 | — | — | |
| Ours (mix-AAO)K=16, Base Model=KnowPrompt2025.05 | 35.56 | — | — | |
| RetrievalREK=162025.05 | 35.16 | — | — | |
| CPSupervised training data percentage=1%, Input configuration=OnlyM2020.10 | 35 | — | — | |
| Ours (mix-OBO)K=16, Base Model=KnowPrompt2025.05 | 34.93 | — | — | |
| Ours (mix-AAO)K=8, Base Model=RetrievalRE2025.05 | 34.91 | — | — | |
| Ours (mix-AAO)K=8, Base Model=KnowPrompt2025.05 | 34.3 | — | — | |
| Ours (mix-OBO)K=8, Base Model=RetrievalRE2025.05 | 33.75 | — | — | |
| Ours (mix-OBO)K=8, Base Model=KnowPrompt2025.05 | 33.48 | — | — | |
| RetrievalREK=82025.05 | 32.04 | — | — | |
| MTBSupervised training data percentage=1%, Input configuration=OnlyM2020.10 | 30.8 | — | — | |
| Data GenerationK=16, Base Model=RetrievalRE2025.05 | 30.54 | — | — | |
| Ours (pure)K=32, Base Model=KnowPrompt2025.05 | 30.41 | — | — | |
| MTBSupervised training data percentage=1%, Input configuration=C+M2020.10 | 30.4 | — | — | |
| Ours (pure)K=32, Base Model=RetrievalRE2025.05 | 30.29 | — | — | |
| KnowPromptK=162025.05 | 30 | — | — | |
| Ours (pure)K=16, Base Model=RetrievalRE2025.05 | 29.63 | — | — | |
| Ours (pure)K=16, Base Model=KnowPrompt2025.05 | 27.99 | — | — | |
| Data GenerationK=32, Base Model=KnowPrompt2025.05 | 26.19 | — | — | |
| Data GenerationK=32, Base Model=RetrievalRE2025.05 | 25.99 | — | — | |
| Ours (constant)K=16, Base Model=RetrievalRE2025.05 | 25.31 | — | — | |
| Ours (pure)K=8, Base Model=RetrievalRE2025.05 | 25.2 | — | — | |
| Data GenerationK=8, Base Model=RetrievalRE2025.05 | 24.97 | — | — | |
| Data GenerationK=16, Base Model=KnowPrompt2025.05 | 24.71 | — | — | |
| MTBSupervised training data percentage=1%, Input configuration=OnlyC2020.10 | 23.1 | — | — | |
| Ours (pure)K=8, Base Model=KnowPrompt2025.05 | 22.48 | — | — | |
| Ours (constant)K=8, Base Model=RetrievalRE2025.05 | 22.48 | — | — | |
| KnowPromptK=82025.05 | 22.07 | — | — | |
| BERTSupervised training data percentage=1%, Input configuration=OnlyM2020.10 | 22 | — | — | |
| Ours (constant)K=16, Base Model=KnowPrompt2025.05 | 21.71 | — | — | |
| Data GenerationK=8, Base Model=KnowPrompt2025.05 | 21.67 | — | — | |
| Direct-REPrompting=Direct LLM2025.05 | 21.17 | — | — | |
| BERTSupervised training data percentage=1%, Input configuration=C+M2020.10 | 21.1 | — | — | |
| Ours (constant)K=8, Base Model=KnowPrompt2025.05 | 17.39 | — | — | |
| BERTSupervised training data percentage=1%, Input configuration=OnlyC2020.10 | 16.7 | — | — |