Spatial Reasoning on SpatialEval (test)
35.2Maze Navigation AccIVC-Prune
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| IVC-PruneModels=Qwen2.5-VL 7B, Average Tokens=50%2026.02 | 35.2 | — | 83.4 | 69.9 | 70.4 | 63 | |
| FastVModels=Qwen2.5-VL 7B, Average Tokens=54%2026.02 | 33.5 | — | 78.3 | 68.1 | 69.6 | 60.2 | |
| VanillaModels=InternVL 2.5 8B, Average Tokens=100%2026.02 | 33.5 | — | 76.2 | 63.5 | 65.9 | 58 | |
| PDropModels=DeepSeek-VL2 Small-16B, Average Tokens=57%2026.02 | 33.5 | — | 73.8 | 59.1 | 61.5 | 55.6 | |
| VanillaModels=Qwen2.5-VL 7B, Average Tokens=100%2026.02 | 32.8 | — | 84.1 | 68.3 | 70.4 | 62 | |
| PDropModels=InternVL 2.5 8B, Average Tokens=56%2026.02 | 32.8 | — | 66.1 | 60.8 | 61.5 | 53.5 | |
| IVC-PruneModels=InternVL 2.5 8B, Average Tokens=50%2026.02 | 32.7 | — | 76.1 | 63.8 | 68.1 | 57.8 | |
| FastVModels=DeepSeek-VL2 Small-16B, Average Tokens=54%2026.02 | 32.6 | — | 65.9 | 58 | 58.5 | 52.3 | |
| IVC-PruneModels=DeepSeek-VL2 Small-16B, Average Tokens=52%2026.02 | 32.3 | — | 75.3 | 59.3 | 59.3 | 55.7 | |
| VanillaModels=DeepSeek-VL2 Small-16B, Average Tokens=100%2026.02 | 32.1 | — | 75.2 | 59.3 | 60 | 55.6 | |
| VanillaModels=LLaVA-v1.5 7B, Average Tokens=100%2026.02 | 31.3 | — | 30.1 | 43.5 | 38.5 | 35 | |
| PDropModels=LLaVA-v1.5 7B, Average Tokens=47%2026.02 | 30.8 | — | 29.1 | 44 | 35.5 | 34.5 | |
| FastVModels=LLaVA-v1.5 7B, Average Tokens=30%2026.02 | 30.4 | — | 29.5 | 44.1 | 35.5 | 34.5 | |
| IVC-PruneModels=LLaVA-v1.5 7B, Average Tokens=28%2026.02 | 30.3 | — | 30 | 43.1 | 38.5 | 34.6 | |
| FastVModels=InternVL 2.5 8B, Average Tokens=53%2026.02 | 28.1 | — | 54.5 | 52.6 | 60 | 45.5 | |
| PDropModels=Qwen2.5-VL 7B, Average Tokens=61%2026.02 | 26.8 | — | 75.5 | 66.8 | 63.7 | 56.6 |