Aggregated Visual Question Answering Benchmarks (GQA, MMB, MME, POPE, SQA, TextVQA)
64.2GQA ScoreVanilla
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| VanillaBackbone=LLaVA-Next 7B2026.02 | 64.2 | 67.9 | 1,846 | 86.4 | 73.2 | 61.3 | 100 | |
| NuwaBackbone=LLaVA-Next 7B, Average Tokens=6402026.02 | 63.4 | 65.4 | 1,879 | 87.2 | 68.6 | 59.5 | 98.1 | |
| NuwaBackbone=LLaVA-Next 7B, Average Tokens=3202026.02 | 62.3 | 63.2 | 1,813 | 86 | 68.2 | 58.5 | 96.1 | |
| VisionZipBackbone=LLaVA-Next 7B, Average Tokens=6402026.02 | 61.3 | 66.2 | 1,787 | 85.9 | 68.1 | 60.2 | 96.9 | |
| SparseVLMBackbone=LLaVA-Next 7B, Average Tokens=6402026.02 | 60.3 | 65.8 | 1,773 | 84.2 | 67.7 | 57.8 | 95.3 | |
| NuwaBackbone=LLaVA-Next 7B, Average Tokens=1602026.02 | 60 | 60.4 | 1,684 | 83 | 67.5 | 56.3 | 92.29 | |
| VisionZipBackbone=LLaVA-Next 7B, Average Tokens=3202026.02 | 59.2 | 63.1 | 1,702 | 82.1 | 67.3 | 58.9 | 93 | |
| SparseVLMBackbone=LLaVA-Next 7B, Average Tokens=3202026.02 | 57.7 | 63.2 | 1,685 | 82.2 | 67.3 | 55.9 | 91.2 | |
| VisionZipBackbone=LLaVA-Next 7B, Average Tokens=1602026.02 | 55.5 | 60.1 | 1,628 | 74.8 | 68.3 | 56.2 | 87.6 | |
| SparseVLMBackbone=LLaVA-Next 7B, Average Tokens=1602026.02 | 51.2 | 52.1 | 1,542 | 72.7 | 67.5 | 46.4 | 79.8 |