Second-level Implicit Discourse Relation Recognition on CoNLL16 (blind)
66.35AccuracyDiscoPrompt
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DiscoPromptBackbone=T5, Model Size=11b2023.05 | 66.35 | 38.6 | |
| DiscoPromptBackbone=T5, Model Size=large2023.05 | 63.06 | 38.49 | |
| Prompt-TuningBackbone=T5, Model Size=large2023.05 | 57.88 | 35.72 | |
| DiscoPromptBackbone=T5, Model Size=base2023.05 | 57.88 | 39.27 | |
| PCPBackbone=RoBERTa, Model Size=large2023.05 | 56.24 | 27.77 | |
| Prefix-TuningBackbone=T5, Model Size=large2023.05 | 55.53 | 30.53 | |
| BMGF-ROBERTaBackbone=RoBERTa2023.05 | 55.19 | 28.98 | |
| Fine-TuningBackbone=T5, Model Size=large2023.05 | 54.3 | 24.63 | |
| PCPBackbone=RoBERTa, Model Size=base2023.05 | 50.99 | 26 | |
| Fine-TuningBackbone=T5, Model Size=base2023.05 | 49.89 | 24.52 | |
| MTL-Attn-LSTM2023.05 | 40.12 | — | |
| RWP-CNN2023.05 | 39.36 | — | |
| CONLL Baseline2023.05 | 37.67 | — | |
| XLNetBackbone=XLNet, Model Size=large, Case=cased2023.05 | 35.06 | 11.94 | |
| Prompt-TuningBackbone=T5, Model Size=base2023.05 | 33.65 | 11.55 | |
| Prefix-TuningBackbone=T5, Model Size=base2023.05 | 32.71 | 9.7 | |
| XLNetBackbone=XLNet, Model Size=base, Case=cased2023.05 | 28.71 | 9.07 |