Top-level Implicit Discourse Relation Recognition on PDTB 2.0 (Ji split)
75.34F1 ScoreDiscoPrompt (T5-11b)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DiscoPrompt (T5-11b)Backbone=T5-11b2023.05 | 75.34 | 78.06 | |
| DiscoPromptBackbone=T5, Model Scale=11b2023.05 | 75.34 | 78.06 | |
| DiscoPrompt (T5-large)Backbone=T5-large2023.05 | 70.84 | 75.65 | |
| DiscoPromptBackbone=T5, Model Scale=large2023.05 | 70.84 | 75.65 | |
| Long and WebberData Augmentation=true2023.05 | 69.6 | 72.18 | |
| Long and WebberData Augmentation=false2023.05 | 67.85 | 71.7 | |
| ContrastiveIDRR2023.05 | 67.85 | 71.7 | |
| PCP w/ ROBERTa-largeBackbone=RoBERTa-large2023.05 | 67.79 | 73.8 | |
| PCPBackbone=RoBERTa-large2023.05 | 67.79 | 73.8 | |
| Prompt-Tuning (T5-large)Backbone=T5-large, Training Strategy=Prompt-Tuning2023.05 | 66.95 | 71.99 | |
| Prompt-TuningBackbone=T5, Model Scale=large2023.05 | 66.95 | 71.99 | |
| DiscoPrompt (T5-base)Backbone=T5-base2023.05 | 65.79 | 71.7 | |
| DiscoPromptBackbone=T5, Model Scale=base2023.05 | 65.79 | 71.7 | |
| Jiang et al.2023.05 | 65.76 | 72.52 | |
| GOLF2023.05 | 65.76 | 72.52 | |
| Dou et al.2023.05 | 65.06 | 70.17 | |
| PCP w/ ROBERTa-baseBackbone=RoBERTa-base2023.05 | 64.95 | 70.84 | |
| PCPBackbone=RoBERTa-base2023.05 | 64.95 | 70.84 | |
| OTMT (XLNet-large)Backbone=XLNet-large2023.05 | 64.46 | 72.34 | |
| OTMTBackbone=XLNet-large2023.05 | 64.46 | 72.34 | |
| Prefix-Tuning (T5-large)Backbone=T5-large, Training Strategy=Prefix-Tuning2023.05 | 63.74 | 71.51 | |
| Prefix-TuningBackbone=T5, Model Scale=large2023.05 | 63.74 | 71.51 | |
| Wu et al.2023.05 | 63.73 | 71.18 | |
| LDSGM2023.05 | 63.73 | 71.18 | |
| XLNet (large, cased)Backbone=XLNet-large2023.05 | 63.58 | 69.52 | |
| XLNetBackbone=XLNet, Model Scale=large, Casing=cased2023.05 | 63.58 | 69.52 | |
| Liu et al.2023.05 | 63.39 | 69.06 | |
| BMGF-ROBERTaBackbone=RoBERTa2023.05 | 63.39 | 69.06 | |
| Fine-Tuning (T5-large)Backbone=T5-large, Training Strategy=Fine-Tuning2023.05 | 61.37 | 69.69 | |
| Fine-TuningBackbone=T5, Model Scale=large2023.05 | 61.37 | 69.69 | |
| OTMT (XLNet-base)Backbone=XLNet-base2023.05 | 60.78 | 68.89 | |
| OTMTBackbone=XLNet-base2023.05 | 60.78 | 68.89 | |
| XLNet(base, cased)Backbone=XLNet-base2023.05 | 59.33 | 66.35 | |
| XLNetBackbone=XLNet, Model Scale=base, Casing=cased2023.05 | 59.33 | 66.35 | |
| Kurfali and Östling2023.05 | 59.24 | — | |
| Kishimoto et al.2023.05 | 58.48 | 65.26 | |
| Fine-Tuning (T5-base)Backbone=T5-base, Training Strategy=Fine-Tuning2023.05 | 57.61 | 65.39 | |
| Fine-TuningBackbone=T5, Model Scale=base2023.05 | 57.61 | 65.39 | |
| Jiang et al.2023.05 | 57.18 | — | |
| CG-T5Backbone=T52023.05 | 57.18 | 65.54 | |
| BERT-largeBackbone=BERT-large2023.05 | 57.06 | 67.59 | |
| Nguyen et al.2023.05 | 53 | — | |
| MTL-MLoss2023.05 | 53 | — | |
| Dai and Huang2023.05 | 52.89 | 59.66 | |
| ELMo-C&EBackbone=ELMo2023.05 | 52.89 | 59.66 | |
| Bai and Zhao2023.05 | 51.06 | — | |
| Varia et al.2023.05 | 50.2 | 59.13 | |
| RWP-CNN2023.05 | 50.2 | 59.13 | |
| Dai and Huang2023.05 | 48.82 | 57.44 | |
| Guo et al.2023.05 | 47.9 | 57.25 | |
| Lan et al.2023.05 | 47.8 | 57.39 | |
| Shi and Demberg2023.05 | 46.4 | 61.42 | |
| Liu and Li2023.05 | 46.29 | 57.57 | |
| Liu et al.2023.05 | 44.98 | 57.27 | |
| Wu et al.2023.05 | 44.84 | 58.85 | |
| Xu et al.2023.05 | 44.48 | 60.63 | |
| BERT-baseBackbone=BERT-base2023.05 | 43.17 | 62.14 | |
| Prompt-Tuning (T5-base)Backbone=T5-base, Training Strategy=Prompt-Tuning2023.05 | 30.17 | 56.11 | |
| Prompt-TuningBackbone=T5, Model Scale=base2023.05 | 30.17 | 56.11 | |
| Prefix-Tuning (T5-base)Backbone=T5-base, Training Strategy=Prefix-Tuning2023.05 | 25.87 | 52.45 | |
| Prefix-TuningBackbone=T5, Model Scale=base2023.05 | 25.87 | 52.45 |