Character Reordering on Kanbun character reordering dataset
0.944Kendall's TauRoBERTa-classical-chinese-char
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RoBERTa-classical-chinese-charBackbone=RoBERTa, Domain=Classical Chinese, Granularity=Character2023.05 | 0.944 | 78.3 | |
| RoBERTa-chinese-wwm-extBackbone=RoBERTa, Language=Chinese, Training=Whole Word Masking Extended2023.05 | 0.92 | 71.8 | |
| BERT-chineseBackbone=BERT, Language=Chinese2023.05 | 0.917 | 68.9 | |
| BERT-japanese-charBackbone=BERT, Language=Japanese, Granularity=Character2023.05 | 0.898 | 63.7 | |
| RoBERTa-japanese-char-wwmBackbone=RoBERTa, Language=Japanese, Granularity=Character, Training=Whole Word Masking2023.05 | 0.894 | 60 | |
| HumanDescription=Average of three annotators2023.05 | 0.844 | 60.6 | |
| UD-KundokuDescription=Baseline model2023.05 | 0.77 | 40.2 |