Semantic Role Labeling on CoNLL 12
97.2Micro F1DEEPSTRUCT
Evaluation Results
| Method | Links | |
|---|---|---|
| DEEPSTRUCTtraining protocol=multi-task2022.05 | 97.2 | |
| DEEPSTRUCTtraining protocol=w/ finetune2022.05 | 96 | |
| TANL2022.05 | 87.7 | |
| TANLtraining protocol=multitask2022.05 | 87.7 | |
| Full Fine-tuningBase Model=RoBERTa-large2023.03 | 86.5 | |
| BERT SRL2022.05 | 86.5 | |
| Zero-init.Base Model=RoBERTa-large2023.03 | 85.2 | |
| P-tuning V2 (reproduced)Base Model=RoBERTa-large, reproduced=true2023.03 | 84.7 | |
| P-tuning V2Base Model=RoBERTa-large2023.03 | 84.6 | |
| Dep and Span2022.05 | 83.1 | |
| P-tuningBase Model=RoBERTa-large2023.03 | 67.2 |