Machine Translation on WMT En-Ja 22 (COMET, BLEURT)
88.7COMETDUAL-REFLECT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DUAL-REFLECTbackbone=ChatGPT, prompting=zero-shot2024.06 | 88.7 | 67.9 | |
| ChatGPT + MAPSprompting=zero-shot2024.06 | 88.5 | 67.4 | |
| ChatGPT + Refine_cosprompting=zero-shot2024.06 | 88.4 | 66.8 | |
| ChatGPT + Self-Reflectprompting=zero-shot2024.06 | 88.3 | 66.9 | |
| ChatGPTprompting=5-shot2024.06 | 88.2 | 67.1 | |
| ChatGPT + Refineprompting=zero-shot2024.06 | 88.1 | 66.4 | |
| ChatGPT + Rerankprompting=zero-shot2024.06 | 88 | 66.6 | |
| ChatGPTprompting=zero-shot2024.06 | 87.9 | 66.3 | |
| DUAL-REFLECTbackbone=Vicuna-7B, prompting=zero-shot2024.06 | 85.1 | 61.1 | |
| Vicuna-7B + MAPSprompting=zero-shot2024.06 | 84.4 | 60.3 | |
| Vicuna-7Bprompting=5-shot2024.06 | 83.3 | 59.3 | |
| Vicuna-7Bprompting=zero-shot2024.06 | 82.3 | 58.7 | |
| DUAL-REFLECTbackbone=Alpaca-7B, prompting=zero-shot2024.06 | 61 | 34.7 | |
| Alpaca-7B + MAPSprompting=zero-shot2024.06 | 58.2 | 33.9 | |
| Alpaca-7Bprompting=5-shot2024.06 | 57.9 | 31.9 | |
| Alpaca-7Bprompting=zero-shot2024.06 | 56.6 | 31.4 |