Fine-grained Video Understanding on VCG-Bench
3.32Consistency ScoreVanilla
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| VanillaRetention Ratio=100%2026.04 | 3.32 | 2.3 | 3.28 | 2.38 | 2.98 | — | |
| Holitom+SToPRetention Ratio=20%2026.04 | 3.23 | 2.18 | 3.21 | 2.23 | 2.96 | 1.84 | |
| HolitomRetention Ratio=20%2026.04 | 3.22 | 2.15 | 3.2 | 2.22 | 2.95 | 2.41 | |
| VisionZipRetention Ratio=20%2026.04 | 3.22 | 2.17 | 3.14 | 2.14 | 2.91 | 4.85 | |
| HolitomRetention Ratio=15%2026.04 | 3.18 | 2.15 | 3.17 | 2.16 | 2.92 | 3.52 | |
| PruneVidRetention Ratio=20%2026.04 | 3.16 | 2.15 | 3.12 | 2.14 | 2.86 | 5.74 | |
| VisionZip+SToPRetention Ratio=20%2026.04 | 3.16 | 2.09 | 3.17 | 2.19 | 2.89 | 3.66 | |
| VisionZip+SToPRetention Ratio=15%2026.04 | 3.16 | 2.12 | 3.12 | 2.15 | 2.87 | 4.36 | |
| FastVidRetention Ratio=20%2026.04 | 3.15 | 2.11 | 3.16 | 2.16 | 2.9 | 8.21 | |
| FastVid+SToPRetention Ratio=20%2026.04 | 3.15 | 2.15 | 3.21 | 2.24 | 2.96 | 2.61 | |
| Holitom+SToPRetention Ratio=15%2026.04 | 3.15 | 2.16 | 3.18 | 2.2 | 2.96 | 2.78 | |
| FastVid+SToPRetention Ratio=15%2026.04 | 3.15 | 2.09 | 3.14 | 2.17 | 2.88 | 3.42 | |
| PruneVidRetention Ratio=15%2026.04 | 3.14 | 2.17 | 3.08 | 2.11 | 2.87 | 5.68 | |
| FastVidRetention Ratio=15%2026.04 | 3.14 | 2.1 | 3.03 | 2.06 | 2.8 | 12.3 | |
| FastVid+SToPRetention Ratio=10%2026.04 | 3.09 | 2.03 | 3.06 | 2.09 | 2.83 | 6.32 | |
| HolitomRetention Ratio=10%2026.04 | 3.07 | 2.06 | 3.08 | 2.07 | 2.84 | 6.22 | |
| Holitom+SToPRetention Ratio=10%2026.04 | 3.07 | 2.03 | 3.1 | 2.11 | 2.84 | 4.8 | |
| VisionZipRetention Ratio=15%2026.04 | 3.05 | 2.1 | 3.03 | 2.05 | 2.78 | 7.87 | |
| VisionZip+SToPRetention Ratio=10%2026.04 | 3.05 | 2.06 | 3.05 | 2.08 | 2.8 | 6.87 | |
| PruneVidRetention Ratio=10%2026.04 | 3.02 | 2.04 | 3.02 | 2.04 | 2.78 | 9.22 | |
| FastVidRetention Ratio=10%2026.04 | 2.95 | 1.95 | 2.92 | 1.94 | 2.72 | 15.69 | |
| VisionZipRetention Ratio=10%2026.04 | 2.81 | 1.84 | 2.78 | 1.84 | 2.55 | 16.79 |