Visual Question Answering on VQA v2 (Absolute Score, Retention Rate)
78.5Absolute ScoreVanilla
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| VanillaBackbone=LLaVA-1.5-7B, Retained Tokens=5762026.04 | 78.5 | 100 | 100 | |
| RCPBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 77.54 | 98.78 | 99.06 | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 76.9 | 97.96 | 98.75 | |
| DARTBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 76.8 | 97.83 | 98.57 | |
| RCPBackbone=LLaVA-1.5-7B, Retained Tokens=1282026.04 | 76.8 | 97.83 | 96.73 | |
| PDropBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 75.2 | 95.79 | 97.07 | |
| HiREDBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 75 | 95.54 | 96.53 | |
| RCPBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 74.53 | 94.94 | 95 | |
| HiREDBackbone=LLaVA-1.5-7B, Retained Tokens=1282026.04 | 73.49 | 93.62 | 95.2 | |
| PDropBackbone=LLaVA-1.5-7B, Retained Tokens=1282026.04 | 72.99 | 92.98 | 95.08 | |
| DARTBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 72.49 | 92.35 | 94.4 | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 72.4 | 92.23 | 94.81 | |
| HiREDBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 69.79 | 88.9 | 91.46 | |
| PDropBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 69.29 | 88.27 | 75.89 | |
| ToMeBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 68 | 86.62 | 88.32 | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=1922026.04 | 67.1 | 85.48 | 88.71 | |
| ToMeBackbone=LLaVA-1.5-7B, Retained Tokens=1282026.04 | 63 | 80.25 | 82.75 | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=1282026.04 | 61.8 | 78.73 | 82.86 | |
| ToMeBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 57.1 | 72.74 | 73.34 | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=642026.04 | 55 | 70.06 | 73.88 |