Real-world Multimodal Evaluation on MME-RW (test)
62.9Overall ScoreQwen2.5-VL-7B + Saliency-R1
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-7B + Saliency-R1Training Stage=Saliency-R12026.04 | 62.9 | |
| Qwen2.5-VL-7B + SFTTraining Stage=SFT2026.04 | 60.6 | |
| Qwen2.5-VL-7BTraining Stage=Base2026.04 | 58.7 | |
| InternVL2.5-8B2026.04 | 58.1 | |
| Qwen2.5-VL-3B + Saliency-R1Training Stage=Saliency-R12026.04 | 57.7 | |
| LLaVA-OneVision-7B2026.04 | 57.4 | |
| Qwen2.5-VL-3B + SFTTraining Stage=SFT2026.04 | 56 | |
| InternVL2-8B2026.04 | 53.5 | |
| Qwen2.5-VL-3BTraining Stage=Base2026.04 | 52 | |
| Claude-3.5 Sonnet2026.04 | 51.6 | |
| IXC-2.52026.04 | 50 | |
| MiniCPM-Llama-V-2.5-8B2026.04 | 45.6 | |
| GPT-4o2026.04 | 45.2 | |
| Cambrain-1-8B2026.04 | 42.7 | |
| Gemini-1.5-Pro2026.04 | 38.2 |