Machine Translation on Synthetic Image-to-Image Translation Dataset French-English
43.5BLEUGemini-2.5-pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini-2.5-pro2026.05 | 43.5 | 62.6 | 43.1 | |
| Gemini-2.5-flash2026.05 | 42.2 | 61.1 | 45.4 | |
| Gemini-2.5-lite2026.05 | 39.5 | 59.2 | 46.6 | |
| EuroLLMParameters=9B2026.05 | 37.6 | 59 | 45.3 | |
| EuroLLMParameters=1.7B2026.05 | 35.7 | 57.2 | 49.1 | |
| Llama-3.1Parameters=8B2026.05 | 35.3 | 57.1 | 48 | |
| Translatotron-VSource=Private communication2026.05 | 34.2 | — | — | |
| Llama-3.2Parameters=3B2026.05 | 32.9 | 55 | 52.9 | |
| Llama-3.2Parameters=1B2026.05 | 32.3 | 53.7 | 54.8 | |
| M2M100Parameters=1.2B2026.05 | 25.2 | 46.1 | 65.3 |