Accuracy on Visual Question Answering on VQA v2
81.3AccuracyLLaVA-NeXT-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA-NeXT-7BRetained Tokens=2,880, Token Reduction Ratio=100%2026.07 | 81.3 | |
| VILA-13BModel Scale=13B, Precision=FP162023.06 | 80.5 | |
| VILA-13B-AWQModel Scale=13B, Quantization=INT4-g1282023.06 | 80.4 | |
| VILA-7BModel Scale=7B, Precision=FP162023.06 | 80.3 | |
| VILA-7B-AWQModel Scale=7B, Quantization=INT4-g1282023.06 | 80.1 | |
| EADPRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 80 | |
| CDPrunerRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 79.9 | |
| SparseVLMRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 79.5 | |
| DivPruneRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 79.3 | |
| VisionZipRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 79.2 | |
| HiPruneRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 78.8 | |
| PDropRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 78.7 | |
| DARTRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 78.7 | |
| EADPRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 78.6 | |
| PruMerge+Retained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 78 | |
| TRIMRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 77.9 | |
| CDPrunerRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 77.9 | |
| DivPruneRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 77.2 | |
| EADPRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 77 | |
| FastVRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 76.6 | |
| CDPrunerRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 76.4 | |
| VisionZipRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 76.3 | |
| TRIMRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 75.3 | |
| DARTRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 75.1 | |
| DivPruneRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 75 | |
| PruMerge+Retained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 74.8 | |
| HiPruneRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 74.4 | |
| SparseVLMRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 73.9 | |
| DARTRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 73.1 | |
| VisionZipRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 70.8 | |
| TRIMRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 70.3 | |
| PruMerge+Retained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 69.1 | |
| PDropRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 67.7 | |
| HiPruneRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 67.3 | |
| FastVRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 63.4 |