Machine Translation on WMT22 Sah-Ru (test)
59.5COMET ScoreDUAL-REFLECT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DUAL-REFLECTInference Methodology=DUAL-REFLECT2024.06 | 59.5 | 37.9 | — | |
| IBUTMode=Iterative Bilingual Understanding (IBUT)2024.10 | 59.5 | 37.9 | 6.9 | |
| ChatGPTMode=Multi-Agent Prompt Selection (MAPS)2024.10 | 58.7 | 37.3 | 6.7 | |
| ChatGPTMode=Reranking2024.10 | 58.6 | 36.3 | 6.5 | |
| ChatGPT + 5-shotPrompting Strategy=Few-shot, Shots=52024.06 | 58.3 | 36 | — | |
| ChatGPTMode=5-shot prompting2024.10 | 58.3 | 36 | 6.4 | |
| ChatGPTMode=Iterative Refinement2024.10 | 58.3 | 37.4 | 6.7 | |
| ChatGPTMode=Translation Error Analysis and Reduction (TEaR)2024.10 | 58.3 | 37.2 | 6.7 | |
| ChatGPT + MADInference Methodology=MAD2024.06 | 58.1 | 37.1 | — | |
| ChatGPTMode=Multi-Agent Discussion (MAD)2024.10 | 58.1 | 37.1 | 6.7 | |
| ChatGPTMode=Dual-Reflection2024.10 | 58 | 37.1 | 6.5 | |
| ChatGPTPrompting Strategy=Zero-shot2024.06 | 57.5 | 36 | — | |
| ChatGPTMode=Zero-shot2024.10 | 57.5 | 36 | 5.9 |