General Multimodal Understanding on Combined 9 Benchmarks
100Average AccuracyLLaVA-NeXT Vanilla 13B
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA-NeXT Vanilla 13BToken Budget=28802025.08 | 100 | |
| VisionZip ✨Token Budget=6402025.08 | 98.8 | |
| MMTokToken Budget=6402025.08 | 98.2 | |
| VisionZip ✨Token Budget=3202025.08 | 97.8 | |
| VisionZipToken Budget=6402025.08 | 97.7 | |
| DivPruneToken Budget=6402025.08 | 97.1 | |
| MMTokToken Budget=3202025.08 | 96.4 | |
| MMTokToken Budget=1602025.08 | 95.1 | |
| VisionZipToken Budget=3202025.08 | 94.7 | |
| VisionZip ✨Token Budget=1602025.08 | 94.6 | |
| DivPruneToken Budget=3202025.08 | 94.5 | |
| DivPruneToken Budget=1602025.08 | 92 | |
| VisionZipToken Budget=1602025.08 | 91.4 |