Text-Image Machine Translation on MMTIT-Bench other2en 1.0 (test)
83.98Gemini-flash Judge ScoreQwen2.5-VLsft(CPR-Trans)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Qwen2.5-VLsft(CPR-Trans)Param.=7B, Think=true2026.03 | 83.98 | 82.99 | 81.07 | |
| Gemini2.5-flashParam.=-, Think=true2026.03 | 82.94 | 74.76 | 79.26 | |
| dots.vlm1Param.=671B-A37B, Think=true2026.03 | 79.23 | 73.82 | 76.84 | |
| Seed1.6-ThinkingParam.=-, Think=true2026.03 | 74.09 | 73.72 | 76.23 | |
| Qwen3-VL-ThinkingParam.=235B-A22B, Think=true2026.03 | 73.81 | 70.38 | 76.21 | |
| Qwen2.5-VLsft(Direct)Param.=7B, Think=false2026.03 | 68.4 | 69.04 | 75.42 | |
| MiMo-VLParam.=7B, Think=true2026.03 | 67.72 | 64.81 | 74.96 | |
| Seed1.6Param.=-, Think=false2026.03 | 67.21 | 62.61 | 74.01 | |
| Dots.ocr+Qwen3Param.=-, Think=false2026.03 | 66.24 | 61.2 | 73.62 | |
| Qwen3-VL-InstructParam.=235B-A22B, Think=false2026.03 | 64.39 | 61.75 | 73.67 | |
| Qwen2.5-VL_OriginParam.=7B, Think=false2026.03 | 53.98 | 52.03 | 70.27 | |
| MinerU2.5+Qwen3Param.=-, Think=false2026.03 | 48.32 | 45.13 | 63.43 |