Machine Translation on FLORES-200 snd_Arab-English (test)
87.5COMET-22SCALE-refine
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SCALE-refineLLM=GPT-4, STM=NLLB-3.3B2023.09 | 87.5 | 79.5 | 76.6 | 42.9 | |
| NLLBModel Version=3.3B, Evaluation Protocol=Supervised2023.09 | 86.9 | 79.5 | 75.5 | 44.4 | |
| MicrosoftType=Commercial System2023.09 | 83.6 | 77.4 | 70.4 | 38.5 | |
| GPT-4Evaluation Protocol=few-shot2023.09 | 83.2 | 75.3 | 69.9 | 32.3 | |
| GPT-3.5Evaluation Protocol=few-shot2023.09 | 75.6 | 68.1 | 58.8 | 17.3 | |
| XGLMEvaluation Protocol=few-shot2023.09 | 53.8 | 45.1 | 29.8 | 1.8 | |
| M2M100Type=Specialized Model2023.09 | 49.8 | 47.2 | 39.2 | 6.4 |