Visual Reasoning on GQA (test)
69AccuracyRandOpt
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RandOptModel=Qwen2.5-VL-3B-Inst, N=5000, K=502026.03 | 69 | — | |
| InternVL-2-26BModel=InternVL-2-26B2026.01 | 65.1 | — | |
| InternVL-2-8BModel=InternVL-2-8B2026.01 | 62.7 | — | |
| Qwen2.5-VL-7BPruning ratio=0%2026.03 | 62.2 | 100 | |
| ResPrunePruning ratio=66.7%2026.03 | 59.9 | 98.4 | |
| LLaVA-OV-7B w/ CLIModel=LLaVA-OV-7B w/ CLI2026.01 | 59.7 | — | |
| LLaVA-OV-7BModel=LLaVA-OV-7B2026.01 | 59.4 | — | |
| ResPrunePruning ratio=77.8%2026.03 | 59.4 | 96.9 | |
| LLaVA-OV-7B w/ SLIModel=LLaVA-OV-7B w/ SLI2026.01 | 59.3 | — | |
| DivPrunePruning ratio=66.7%2026.03 | 59 | 96.7 | |
| DivPrunePruning ratio=77.8%2026.03 | 58.6 | 95.1 | |
| LLaVA-OV-7B w/ DeepStackModel=LLaVA-OV-7B w/ DeepStack2026.01 | 58.5 | — | |
| FastVPruning ratio=66.7%2026.03 | 58 | 92.3 | |
| IXC-2.5-7BModel=IXC-2.5-7B2026.01 | 57.7 | — | |
| ResPrunePruning ratio=88.9%2026.03 | 57.7 | 92.2 | |
| DivPrunePruning ratio=88.9%2026.03 | 56.9 | 90.4 | |
| FastVPruning ratio=77.8%2026.03 | 56.7 | 88.9 | |
| BaseModel=Qwen2.5-VL-3B-Inst2026.03 | 56.6 | — | |
| LLaVA-OV-0.5B w/ SLIModel=LLaVA-OV-0.5B w/ SLI2026.01 | 54.5 | — | |
| LLaVA-OV-0.5BModel=LLaVA-OV-0.5B2026.01 | 53.7 | — | |
| LLaVA-OV-0.5B w/ CLIModel=LLaVA-OV-0.5B w/ CLI2026.01 | 53.7 | — | |
| LLaVA-OV-0.5B w/ DeepStackModel=LLaVA-OV-0.5B w/ DeepStack2026.01 | 53.3 | — | |
| FastVPruning ratio=88.9%2026.03 | 51.9 | 82.6 | |
| VILA-13BModel=VILA-13B2026.01 | 17.6 | — |