Machine Translation on Synthetic English-French Dataset
44.6BLEU ScoreGemini-2.5-pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini-2.5-proModel Size=pro2026.05 | 44.6 | 65.2 | 44.1 | |
| Gemini-2.5-liteModel Size=lite2026.05 | 40.5 | 62.2 | 47.3 | |
| Gemini-2.5-flashModel Size=flash2026.05 | 37.3 | 57.2 | 55.1 | |
| EuroLLM-9BParameters=9B2026.05 | 36.8 | 59.9 | 50.1 | |
| M2M100-1.2BParameters=1.2B2026.05 | 34.7 | 57.7 | 55 | |
| EuroLLM-1.7BParameters=1.7B2026.05 | 33.6 | 56 | 55.7 | |
| Llama-3.1-8BParameters=8B2026.05 | 29.3 | 55.5 | 55.8 | |
| Llama-3.2-3BParameters=3B2026.05 | 27.5 | 52.5 | 60.1 | |
| Llama-3.2-1BParameters=1B2026.05 | 22.4 | 46.5 | 67.8 |