Visual Question Answering on GQA (Acc., Relative Performance)
61.9AccuracyLLaVA-1.5 Vanilla 7B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLaVA-1.5 Vanilla 7BVisual Tokens=5762025.08 | 61.9 | 100 | |
| VisionZip 🔥Visual Tokens=1922025.08 | 60.1 | 97.1 | |
| MMTokVisual Tokens=1922025.08 | 60.07 | 97 | |
| DivPruneVisual Tokens=1922025.08 | 59.97 | 96.9 | |
| VisionZipVisual Tokens=1922025.08 | 59.3 | 95.8 | |
| MMTokVisual Tokens=1282025.08 | 59.29 | 95.8 | |
| DivPruneVisual Tokens=1282025.08 | 59.25 | 95.7 | |
| VisionZip 🔥Visual Tokens=1282025.08 | 58.9 | 95.2 | |
| MMTokVisual Tokens=642025.08 | 58.29 | 94.2 | |
| DivPruneVisual Tokens=642025.08 | 57.78 | 93.3 | |
| SparseVLMVisual Tokens=1922025.08 | 57.6 | 93.1 | |
| VisionZipVisual Tokens=1282025.08 | 57.6 | 93.1 | |
| VisionZip 🔥Visual Tokens=642025.08 | 57 | 92.1 | |
| SparseVLMVisual Tokens=1282025.08 | 56 | 90.5 | |
| VisionZipVisual Tokens=642025.08 | 55.1 | 89 | |
| FastVVisual Tokens=1922025.08 | 52.7 | 85.1 | |
| SparseVLMVisual Tokens=642025.08 | 52.7 | 85.1 | |
| FastVVisual Tokens=1282025.08 | 49.6 | 80.1 | |
| FastVVisual Tokens=642025.08 | 46.1 | 74.5 |