Action-Conditioned Object Grounded Reasoning on Know-Show (test)
89.3Task AccuracyHuman
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| HumanSubset=Know-Show subset2025.12 | 89.3 | 100 | 89.3 | 96 | 90.7 | |
| Gemini-2.5-ProSubset=Know-Show subset2025.12 | 72 | 0 | 0 | 0 | 0 | |
| GRAM2025.12 | 50.4 | 46 | 27.8 | 18 | 10.8 | |
| Ovis2.5-9B2025.12 | 49.6 | 10.6 | 5.8 | 1.6 | 1.2 | |
| Gemini-2.5-FlashSubset=Know-Show subset2025.12 | 48 | 0 | 0 | 0 | 0 | |
| MiniCPM-V 4.52025.12 | 40.4 | 1.4 | 0.4 | 0.2 | 0 | |
| LLaVA-OneVision2025.12 | 31.2 | 14 | 4.6 | 1.4 | 0.8 | |
| GPT-4oSubset=Know-Show subset2025.12 | 28 | 0 | 0 | 0 | 0 | |
| Qwen2.5 VL 7B2025.12 | 24 | 5.6 | 2.2 | 2.2 | 1 | |
| VideoLLaVA2025.12 | 1.2 | 23.8 | 0.2 | 6.6 | 0.2 |