Visual Question Answering on VQA v2 (Accuracy & Retention %)
81.2AccuracyLLaVA-NeXT-7B (Upper Bound)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LLaVA-NeXT-7B (Upper Bound)Backbone=LLaVA-NeXT-7B, Tokens Kept=2880, Pruning Ratio=0.0%2025.12 | 81.2 | 100 | |
| FlashVLMBackbone=LLaVA-NeXT-7B, Tokens Kept=640, Pruning Ratio=77.8%2025.12 | 80.5 | 99.12 | |
| VisPrunerBackbone=LLaVA-NeXT-7B, Tokens Kept=640, Pruning Ratio=77.8%2025.12 | 79.8 | 98.6 | |
| VisionZipBackbone=LLaVA-NeXT-7B, Tokens Kept=640, Pruning Ratio=77.8%2025.12 | 79.2 | 96.7 | |
| FastVBackbone=LLaVA-NeXT-7B, Tokens Kept=640, Pruning Ratio=77.8%2025.12 | 78.9 | 97 | |
| SparseVLMBackbone=LLaVA-NeXT-7B, Tokens Kept=640, Pruning Ratio=77.8%2025.12 | 78.2 | 94.9 | |
| FlashVLMBackbone=LLaVA-NeXT-7B, Tokens Kept=320, Pruning Ratio=88.9%2025.12 | 76.5 | 94.61 | |
| VisPrunerBackbone=LLaVA-NeXT-7B, Tokens Kept=320, Pruning Ratio=88.9%2025.12 | 75.7 | 93.3 | |
| VisionZipBackbone=LLaVA-NeXT-7B, Tokens Kept=320, Pruning Ratio=88.9%2025.12 | 74.2 | 90.5 | |
| FlashVLMBackbone=LLaVA-NeXT-7B, Tokens Kept=160, Pruning Ratio=94.4%2025.12 | 72.2 | 88.03 | |
| FastVBackbone=LLaVA-NeXT-7B, Tokens Kept=320, Pruning Ratio=88.9%2025.12 | 71.9 | 87.7 | |
| SparseVLMBackbone=LLaVA-NeXT-7B, Tokens Kept=320, Pruning Ratio=88.9%2025.12 | 71.4 | 87.9 | |
| VisPrunerBackbone=LLaVA-NeXT-7B, Tokens Kept=160, Pruning Ratio=94.4%2025.12 | 70.6 | 86.7 | |
| VisionZipBackbone=LLaVA-NeXT-7B, Tokens Kept=160, Pruning Ratio=94.4%2025.12 | 67.3 | 82.3 | |
| SparseVLMBackbone=LLaVA-NeXT-7B, Tokens Kept=160, Pruning Ratio=94.4%2025.12 | 62.2 | 74.5 | |
| FastVBackbone=LLaVA-NeXT-7B, Tokens Kept=160, Pruning Ratio=94.4%2025.12 | 61.8 | 74.7 |