Multimodal Model Evaluation on MME (test)
2,310MME ScoreQwen2.5-VL-7B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B, Token Retention Ratio=Full Tokens2026.06 | 2,310 | 100 | |
| VisionZipBackbone=Qwen2.5-VL-7B, Token Retention Ratio=20%2026.06 | 2,221 | 95.6 | |
| STSBackbone=Qwen2.5-VL-7B, Token Retention Ratio=20%2026.06 | 2,211 | 97.1 | |
| DivPruneBackbone=Qwen2.5-VL-7B, Token Retention Ratio=20%2026.06 | 2,173 | 96 | |
| STSBackbone=Qwen2.5-VL-7B, Token Retention Ratio=10%2026.06 | 2,108 | 92.2 | |
| DivPruneBackbone=Qwen2.5-VL-7B, Token Retention Ratio=10%2026.06 | 2,054 | 90.5 | |
| VisionZipBackbone=Qwen2.5-VL-7B, Token Retention Ratio=10%2026.06 | 1,937 | 88.7 |