Second-level Implicit Discourse Relation Recognition on PDTB 2.0 (Ji split)
68.14AccuracyDiscoPrompt (T5-11b)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DiscoPrompt (T5-11b)Backbone=T5-11b2023.05 | 68.14 | 52.42 | |
| DiscoPromptBackbone=T5, Model Scale=11b2023.05 | 68.14 | 52.42 | |
| DiscoPrompt (T5-large)Backbone=T5-large2023.05 | 64.58 | 49.03 | |
| DiscoPromptBackbone=T5, Model Scale=large2023.05 | 64.58 | 49.03 | |
| Long and WebberData Augmentation=true2023.05 | 61.69 | 49.66 | |
| PCP w/ ROBERTa-largeBackbone=RoBERTa-large2023.05 | 61.41 | 44.04 | |
| PCPBackbone=RoBERTa-large2023.05 | 61.41 | 44.04 | |
| XLNet (large, cased)Backbone=XLNet-large2023.05 | 61.29 | 38.24 | |
| XLNetBackbone=XLNet, Model Scale=large, Casing=cased2023.05 | 61.29 | 38.24 | |
| Jiang et al.2023.05 | 61.16 | 41.74 | |
| GOLF2023.05 | 61.16 | 41.74 | |
| OTMT (XLNet-large)Backbone=XLNet-large2023.05 | 61.06 | — | |
| OTMTBackbone=XLNet-large2023.05 | 61.06 | — | |
| DiscoPrompt (T5-base)Backbone=T5-base2023.05 | 61.02 | 43.68 | |
| DiscoPromptBackbone=T5, Model Scale=base2023.05 | 61.02 | 43.68 | |
| PCP w/ ROBERTa-baseBackbone=RoBERTa-base2023.05 | 60.54 | 41.55 | |
| PCPBackbone=RoBERTa-base2023.05 | 60.54 | 41.55 | |
| Wu et al.2023.05 | 60.33 | 40.49 | |
| LDSGM2023.05 | 60.33 | 40.49 | |
| Prompt-Tuning (T5-large)Backbone=T5-large, Training Strategy=Prompt-Tuning2023.05 | 60.15 | 44.08 | |
| Prompt-TuningBackbone=T5, Model Scale=large2023.05 | 60.15 | 44.08 | |
| Prefix-Tuning (T5-large)Backbone=T5-large, Training Strategy=Prefix-Tuning2023.05 | 59.77 | 39.73 | |
| Prefix-TuningBackbone=T5, Model Scale=large2023.05 | 59.77 | 39.73 | |
| Long and WebberData Augmentation=false2023.05 | 59.19 | 45.54 | |
| ContrastiveIDRR2023.05 | 59.19 | 45.54 | |
| Liu et al.2023.05 | 58.13 | 35.25 | |
| BMGF-ROBERTaBackbone=RoBERTa2023.05 | 58.13 | 35.25 | |
| Fine-Tuning (T5-large)Backbone=T5-large, Training Strategy=Fine-Tuning2023.05 | 57.65 | 38.04 | |
| Fine-TuningBackbone=T5, Model Scale=large2023.05 | 57.65 | 38.04 | |
| OTMT (XLNet-base)Backbone=XLNet-base2023.05 | 56.65 | — | |
| OTMTBackbone=XLNet-base2023.05 | 56.65 | — | |
| Fine-Tuning (T5-base)Backbone=T5-base, Training Strategy=Fine-Tuning2023.05 | 55.53 | 33.96 | |
| Fine-TuningBackbone=T5, Model Scale=base2023.05 | 55.53 | 33.96 | |
| Kurfali and Östling2023.05 | 55.42 | 39.33 | |
| XLNet(base, cased)Backbone=XLNet-base2023.05 | 54.73 | 36.36 | |
| XLNetBackbone=XLNet, Model Scale=base, Casing=cased2023.05 | 54.73 | 36.36 | |
| BERT-largeBackbone=BERT-large2023.05 | 54.57 | 30.02 | |
| Kishimoto et al.2023.05 | 54.32 | — | |
| Shi and Demberg2023.05 | 53.23 | — | |
| CG-T5Backbone=T52023.05 | 53.13 | 37.76 | |
| BERT-baseBackbone=BERT-base2023.05 | 50.24 | 26.32 | |
| Nguyen et al.2023.05 | 49.95 | — | |
| MTL-MLoss2023.05 | 49.95 | — | |
| Dai and Huang2023.05 | 48.23 | — | |
| ELMo-C&EBackbone=ELMo2023.05 | 48.23 | 33.41 | |
| Bai and Zhao2023.05 | 48.22 | — | |
| Shi and Demberg2023.05 | 47.83 | — | |
| Qin et al.2023.05 | 46.23 | — | |
| Qin et al.2023.05 | 45.04 | — | |
| Ji and Eisenstein2023.05 | 44.59 | — | |
| Prompt-Tuning (T5-base)Backbone=T5-base, Training Strategy=Prompt-Tuning2023.05 | 38.21 | 15.01 | |
| Prompt-TuningBackbone=T5, Model Scale=base2023.05 | 38.21 | 15.01 | |
| Prefix-Tuning (T5-base)Backbone=T5-base, Training Strategy=Prefix-Tuning2023.05 | 31.09 | 7.49 | |
| Prefix-TuningBackbone=T5, Model Scale=base2023.05 | 31.09 | 7.49 | |
| Jiang et al.2023.05 | — | 37.76 |