Machine Translation on Italian-Korean
18.31BLEUPIVOTE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| PIVOTEMerging Module=GPT-4o, Candidates=Direct & English Pivot2025.02 | 18.31 | 24.32 | 88.33 | |
| PIVOTEMerging Module=GPT-4, Candidates=Direct & English Pivot2025.02 | 18.18 | 24.05 | 88.74 | |
| PIVOTEMerging Module=GPT-4o, Candidates=top32025.02 | 18.08 | 23.98 | 88.15 | |
| PIVOTEMerging Module=GPT-4, Candidates=top32025.02 | 17.95 | 23.84 | 87.5 | |
| NLLBtype=Standalone NMT System2025.02 | 17.4 | 23.39 | 87.33 | |
| GPT-4type=Standalone NMT System2025.02 | 17.23 | 22.96 | 86.94 | |
| GPT-4otype=Standalone NMT System2025.02 | 17.2 | 22.82 | 85.31 | |
| MBRtype=Prior Ensemble Method2025.02 | 17.14 | 23 | 87.53 | |
| PIVOTEMerging Module=Llama-3, Candidates=Direct & English Pivot2025.02 | 14.97 | 20.21 | 85.42 | |
| PIVOTEMerging Module=Llama-3, Candidates=top32025.02 | 14.56 | 19.92 | 87.34 | |
| Llama-3type=Standalone NMT System2025.02 | 13.82 | 18.95 | 85.8 | |
| Vicunatype=Standalone NMT System2025.02 | 10.6 | 16.51 | 72.29 | |
| Baizetype=Standalone NMT System2025.02 | 10.38 | 16.44 | 76.63 | |
| EVAtype=Prior Ensemble Method2025.02 | 1.51 | 3.57 | 37.17 | |
| LLM-Blendertype=Prior Ensemble Method2025.02 | 0.03 | 0.85 | 42.77 |