Machine Translation on WMT De-En 22 (test)
88.14COMETM-DocRefine_doc+sent
Evaluation results
| Method | Links | ||
|---|---|---|---|
| M-DocRefine_doc+sentBase Translation=GPT-4o-mini, Refiner=Mistral-Nemo-Instruct, Fine-tuned=true2025.04 | 88.14 | — | |
| M-DocRefine_doc+sentBase Translation=NLLB, Refiner=Mistral-Nemo-Instruct, Fine-tuned=true2025.04 | 88.1 | — | |
| L-DocRefine_doc+sentBase Translation=GPT-4o-mini, Refiner=LLaMA-3-8B-Instruct, Fine-tuned=true2025.04 | 87.88 | — | |
| L-DocRefine_doc+sentBase Translation=NLLB, Refiner=LLaMA-3-8B-Instruct, Fine-tuned=true2025.04 | 87.85 | — | |
| GPT DocRefine_doc+sentBase Translation=GPT-4o-mini, Refiner=GPT-4o-mini, Fine-tuned=false2025.04 | 87.39 | — | |
| GPT DocRefine_docBase Translation=GPT-4o-mini, Refiner=GPT-4o-mini, Fine-tuned=false2025.04 | 87.04 | — | |
| GPT DocRefine_sentBase Translation=GPT-4o-mini, Refiner=GPT-4o-mini, Fine-tuned=false2025.04 | 87.03 | — | |
| GPT Doc2DocBase Translation=GPT-4o-mini, Refiner=None, Fine-tuned=false2025.04 | 87 | — | |
| GPT SentRefine_sentBase Translation=GPT-4o-mini, Refiner=GPT-4o-mini, Fine-tuned=false2025.04 | 86.86 | — | |
| NLLB Sent2SentBase Translation=NLLB, Refiner=None, Fine-tuned=false2025.04 | 86.79 | — | |
| GPT Sent2SentBase Translation=GPT-4o-mini, Refiner=None, Fine-tuned=false2025.04 | 86.49 | — | |
| WMT22 Winners2024.06 | 85 | 33.7 | |
| TASTEBackbone=LLaMA-2-7b, Training Strategy=Tuning with Fixed Embedding Layer, Subtask Form=Text Classification2024.06 | 84.11 | 31.03 | |
| TASTEBackbone=LLaMA-2-7b, Training Strategy=Tuning with Fixed Embedding Layer, Subtask Form=Quality Estimation2024.06 | 84.07 | 30.75 | |
| MT-FixEmbBackbone=LLaMA-2-7b, Training Strategy=Tuning with Fixed Embedding Layer2024.06 | 84.05 | 30.62 | |
| TASTEBackbone=LLaMA-2-7b, Training Strategy=Full-Parameter Tuning, Subtask Form=Text Classification2024.06 | 83.92 | 30.17 | |
| TASTEBackbone=LLaMA-2-7b, Training Strategy=Full-Parameter Tuning, Subtask Form=Quality Estimation2024.06 | 83.83 | 30.49 | |
| MT-FullBackbone=LLaMA-2-7b, Training Strategy=Full-Parameter Tuning2024.06 | 83.79 | 30.1 | |
| NLLB-3.3b2024.06 | 83.42 | 29.54 | |
| BaylingBackbone=LLaMA-2-7b2024.06 | 83.19 | 28.16 | |
| ParroTBackbone=LLaMA-2-7b2024.06 | 82.4 | 27.3 | |
| TASTEBackbone=BLOOMZ-7b1-mt, Training Strategy=Tuning with Fixed Embedding Layer, Subtask Form=Quality Estimation2024.06 | 79.2 | 26.28 | |
| TASTEBackbone=BLOOMZ-7b1-mt, Training Strategy=Tuning with Fixed Embedding Layer, Subtask Form=Text Classification2024.06 | 78.96 | 26.27 | |
| TIMBackbone=BLOOMZ-7b1-mt2024.06 | 78.94 | 26.12 | |
| TASTEBackbone=BLOOMZ-7b1-mt, Training Strategy=Full-Parameter Tuning, Subtask Form=Text Classification2024.06 | 78.74 | 24.97 | |
| TASTEBackbone=BLOOMZ-7b1-mt, Training Strategy=Full-Parameter Tuning, Subtask Form=Quality Estimation2024.06 | 78.42 | 24.87 | |
| MT-FixEmbBackbone=BLOOMZ-7b1-mt, Training Strategy=Tuning with Fixed Embedding Layer2024.06 | 78.27 | 25.07 | |
| ParroTBackbone=BLOOMZ-7b1-mt2024.06 | 78 | 24.9 | |
| MT-FullBackbone=BLOOMZ-7b1-mt, Training Strategy=Full-Parameter Tuning2024.06 | 77.61 | 24.05 |