Word-level Fine-grained Quality Estimation on en-ru
0.3583MCCRIEQE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RIEQECategory=Our Method2026.05 | 0.3583 | 40.92 | 33.25 | 53.18 | |
| DCSQECategory=Baselines2026.05 | 0.351 | 37.33 | 29.84 | 49.85 | |
| RIEQE-NonThinkingCategory=Our Method, Thinking=false2026.05 | 0.3434 | 38.71 | 31.25 | 50.86 | |
| xComet-XXLCategory=Baselines2026.05 | 0.263 | 32.86 | 27.58 | 40.67 | |
| Qwen3-4B-Thinking-2507Category=Backbone Model, Subtask=true2026.05 | 0.2508 | 31.2 | 19.59 | 50.45 | |
| GPT-5.5Category=Baselines2026.05 | 0.2038 | 25.41 | 15.84 | 64.18 | |
| Qwen3-4B-Thinking-2507Category=Backbone Model2026.05 | 0.1829 | 25.33 | 16.83 | 51.26 |