Second-level Implicit Discourse Relation Recognition on CoNLL16 (test)
69.71AccuracyDiscoPrompt
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DiscoPromptBackbone=T5, Model Size=11b2023.05 | 69.71 | 57.75 | |
| DiscoPromptBackbone=T5, Model Size=large2023.05 | 66.32 | 56.29 | |
| DiscoPromptBackbone=T5, Model Size=base2023.05 | 60.84 | 45.99 | |
| Prompt-TuningBackbone=T5, Model Size=large2023.05 | 60.57 | 49.59 | |
| Prefix-TuningBackbone=T5, Model Size=large2023.05 | 59.4 | 46.06 | |
| Fine-TuningBackbone=T5, Model Size=large2023.05 | 58.88 | 34.66 | |
| PCPBackbone=RoBERTa, Model Size=large2023.05 | 58.36 | 37.79 | |
| BMGF-ROBERTaBackbone=RoBERTa2023.05 | 57.26 | 40.68 | |
| PCPBackbone=RoBERTa, Model Size=base2023.05 | 55.48 | 33.27 | |
| Fine-TuningBackbone=T5, Model Size=base2023.05 | 53.92 | 31.99 | |
| XLNetBackbone=XLNet, Model Size=large, Case=cased2023.05 | 47.85 | 27.13 | |
| MTL-Attn-LSTM2023.05 | 39.4 | — | |
| RWP-CNN2023.05 | 39.39 | — | |
| CONLL Baseline2023.05 | 36.13 | — | |
| XLNetBackbone=XLNet, Model Size=base, Case=cased2023.05 | 33.16 | 18.8 | |
| Prompt-TuningBackbone=T5, Model Size=base2023.05 | 29.11 | 13.01 | |
| Prefix-TuningBackbone=T5, Model Size=base2023.05 | 26.63 | 8.26 |