Machine Translation Quality Estimation on WMT Metrics Shared Task EN–DE 2023
0.83Average CorrelationGEMBA-MQM
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GEMBA-MQM2026.04 | 0.83 | 97 | 0.981 | 57.4 | 0.568 | |
| FairQEBackbone=COMETKiwi 222026.04 | 0.812 | 98.5 | 0.95 | 57.4 | 0.424 | |
| FairQEBackbone=MetricX 24 L2026.04 | 0.806 | 97 | 0.953 | 54.5 | 0.463 | |
| COMETKiwi 23 XL2026.04 | 0.764 | 90.9 | 0.9 | 54.1 | 0.308 | |
| BERTScore2026.04 | 0.754 | 87.9 | 0.891 | 52.8 | 0.325 | |
| MetricX 24 XL2026.04 | 0.75 | 84.8 | 0.872 | 52.8 | 0.378 | |
| COMETKiwi 222026.04 | 0.743 | 86.4 | 0.901 | 54.8 | 0.224 | |
| BLEU2026.04 | 0.742 | 89.4 | 0.917 | 52 | 0.192 | |
| MetricX 24 L2026.04 | 0.734 | 89.4 | 0.94 | 49.3 | 0.155 | |
| chrF2026.04 | 0.722 | 81.8 | 0.866 | 51.9 | 0.232 |