Machine Translation on WMT EN-DE 2022
87.44COMET22GPT-4
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-42025.12 | 87.44 | — | 35.38 | |
| WMT-Best2024.09 | 87.2 | 64.6 | 38.4 | |
| BayLing2024.09 | 86.93 | 62.76 | 34.12 | |
| GPT-3.5-turboprompt=SENSE2024.09 | 86.65 | 62.84 | 34.18 | |
| NLLB-54Bparameters=54B2025.12 | 86.45 | — | 34.5 | |
| DTGshot=5-shot2024.09 | 86.3 | 61.6 | 33.4 | |
| ALMA-13Bparameters=13B2025.12 | 85.62 | — | 31.47 | |
| ALMA-7Bparameters=7B2025.12 | 85.59 | — | 30.31 | |
| GPT-3.5-turboprompt=COT2024.09 | 84.95 | 61.17 | 29.7 | |
| GPT-3.5-turboprompt=Vanilla2024.09 | 84.6 | 60.48 | 33.42 | |
| GPT EVAL2024.09 | 84.2 | 59.6 | 30.9 | |
| Bayling-13Bparameters=13B2025.12 | 82.69 | — | 25.62 | |
| TIM-7Bparameters=7B2025.12 | 82.56 | — | 25.59 | |
| Qwen3-1.7B-SGRPOparameters=1.7B, training=SGRPO2025.12 | 79.13 | — | 21.07 | |
| BigTranslate-13Bparameters=13B2025.12 | 78.81 | — | 21.48 | |
| Qwen3-1.7B-SFTparameters=1.7B, training=SFT2025.12 | 77.7 | — | 19.57 |