Visual Question Answering on ScienceQA (Accuracy and Avg. Ratio)
70.2AccuracyPDrop
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PDropSource=CVPR’25, Average Token=192, Reroute=false2026.06 | 70.2 | 95.7 | |
| PDropSource=CVPR’25, Average Token=128, Reroute=false2026.06 | 69.9 | 93.8 | |
| SparseVLMSource=ICML’25, Average Token=64, Reroute=false2026.06 | 69.8 | 91.5 | |
| VanillaSource=CVPR’24, Average Token=576, Reroute=false2026.06 | 69.5 | 100 | |
| PruMergeSource=ICCV’25, Average Token=64, Reroute=false2026.06 | 69.5 | 92.1 | |
| PDropSource=CVPR’25, Average Token=64, Reroute=false2026.06 | 69.2 | 70.3 | |
| SparseVLMSource=ICML’25, Average Token=192, Reroute=false2026.06 | 69.1 | 95.4 | |
| VisionZipSource=CVPR’25, Average Token=64, Reroute=false2026.06 | 69 | 93.2 | |
| VisionZipSource=CVPR’25, Average Token=192, Reroute=false2026.06 | 68.9 | 97.9 | |
| VisionZipSource=CVPR’25, Average Token=128, Reroute=false2026.06 | 68.9 | 96.9 | |
| Nüwa+RerouteSource=Ours, Average Token=192, Reroute=true2026.06 | 68.6 | 100.6 | |
| NüwaSource=ICLR’26, Average Token=192, Reroute=false2026.06 | 68.2 | 99.2 | |
| Nüwa+RerouteSource=Ours, Average Token=128, Reroute=true2026.06 | 68.2 | 99.8 | |
| NüwaSource=ICLR’26, Average Token=128, Reroute=false2026.06 | 67.8 | 98.4 | |
| Nüwa+RerouteSource=Ours, Average Token=64, Reroute=true2026.06 | 67.7 | 96.7 | |
| PruMergeSource=ICCV’25, Average Token=128, Reroute=false2026.06 | 67.6 | 94.7 | |
| NüwaSource=ICLR’26, Average Token=64, Reroute=false2026.06 | 67.5 | 95.8 | |
| FastVSource=ECCV’24, Average Token=192, Reroute=false2026.06 | 67.3 | 90.1 | |
| SparseVLMSource=ICML’25, Average Token=128, Reroute=false2026.06 | 67.1 | 92.9 | |
| FastVSource=ECCV’24, Average Token=128, Reroute=false2026.06 | 60.2 | 85.2 | |
| FastVSource=ECCV’24, Average Token=64, Reroute=false2026.06 | 51.1 | 77.9 |