Multi-modal Understanding on MMB
67ScoreHiPrune
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HiPruneTokens=640, Retaining Ratio=22.2%, Backbone=LLaVA-NeXT2025.11 | 67 | 101 | |
| VanillaTokens=2880, Retaining Ratio=100%, Backbone=LLaVA-NeXT2025.11 | 66.4 | 100 | |
| VisionZipTokens=640, Retaining Ratio=22.2%, Backbone=LLaVA-NeXT2025.11 | 66.3 | 99.8 | |
| HiPruneTokens=320, Retaining Ratio=11.1%, Backbone=LLaVA-NeXT2025.11 | 65.3 | 98.3 | |
| OC-VTPTokens=640, Retaining Ratio=22.2%, Backbone=LLaVA-NeXT2025.11 | 64.8 | 97.6 | |
| VisionZipTokens=320, Retaining Ratio=11.1%, Backbone=LLaVA-NeXT2025.11 | 63.1 | 95 | |
| OC-VTPTokens=320, Retaining Ratio=11.1%, Backbone=LLaVA-NeXT2025.11 | 61.7 | 92.9 | |
| VisionZipTokens=160, Retaining Ratio=5.6%, Backbone=LLaVA-NeXT2025.11 | 60.1 | 90.5 | |
| OC-VTPTokens=160, Retaining Ratio=5.6%, Backbone=LLaVA-NeXT2025.11 | 60 | 90.4 | |
| HiPruneTokens=160, Retaining Ratio=5.6%, Backbone=LLaVA-NeXT2025.11 | 59.8 | 90.1 |