Detection of omissions on MQM gold dataset ZH-EN
49.6PrecisionSupervised baseline
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Supervised baselineFramework=OpenKiwi, Backbone=XLM-RoBERTa, Tokenizer=Character-level2022.03 | 49.6 | 9.4 | 15.9 | |
| Contrastive Conditioning2022.03 | 25.8 | 62 | 36.5 |