Machine Translation on NH 3.0 (dev test)
35.73BLEUOurs
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ours2026.01 | 35.73 | 52.44 | |
| Azure Translator2026.01 | 31.31 | 49.29 | |
| OursEvaluation protocol=-2026.01 | 18.82 | 44.59 | |
| GPT-5-ReasoningEvaluation Protocol=5-shot2026.01 | 16.33 | 35.75 | |
| Azure TranslatorEvaluation protocol=-2026.01 | 15.14 | 43.76 | |
| GPT-4oEvaluation Protocol=5-shot2026.01 | 13.27 | 31.77 | |
| GPT-5-ReasoningEvaluation Protocol=0-shot2026.01 | 12.8 | 33.1 | |
| GPT-4.1Evaluation Protocol=5-shot2026.01 | 12.14 | 30.76 | |
| GPT-4oEvaluation Protocol=0-shot2026.01 | 9.59 | 28.45 | |
| GPT-4.1Evaluation Protocol=0-shot2026.01 | 8.44 | 27.68 | |
| GPT-5-ReasoningEvaluation protocol=5-shot2026.01 | 6.92 | 29.62 | |
| GPT-5-ReasoningEvaluation protocol=0-shot2026.01 | 4.43 | 23.95 | |
| GPT-4.1Evaluation protocol=5-shot2026.01 | 2.63 | 20.69 | |
| GPT-4oEvaluation protocol=5-shot2026.01 | 1.96 | 16.9 | |
| GPT-4.1Evaluation protocol=0-shot2026.01 | 1.2 | 14.1 | |
| GPT-4oEvaluation protocol=0-shot2026.01 | 1.13 | 12.61 |