Action-Conditioned Temporal Grounded Reasoning on Know-Show 1.0 (test)
94.4Action AccuracyHuman
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| HumanTemporal Window=2 seconds, Subset=Know-Show subset2025.12 | 94.4 | 0.7 | 94.4 | 94.4 | |
| HumanTemporal Window=4 seconds, Subset=Know-Show subset2025.12 | 94.4 | 0.7 | 95.8 | 94.4 | |
| HumanTemporal Window=6 seconds, Subset=Know-Show subset2025.12 | 94.4 | 0.7 | 100 | 94.4 | |
| GPT-4oTemporal Window=2 seconds, Subset=Know-Show subset2025.12 | 64 | 12.3 | 48 | 40 | |
| GPT-4oTemporal Window=4 seconds, Subset=Know-Show subset2025.12 | 64 | 12.3 | 68 | 48 | |
| GPT-4oTemporal Window=6 seconds, Subset=Know-Show subset2025.12 | 64 | 12.3 | 76 | 56 | |
| Gemini-2.5-FlashTemporal Window=2 seconds, Subset=Know-Show subset2025.12 | 64 | 13.8 | 72 | 48 | |
| Gemini-2.5-FlashTemporal Window=4 seconds, Subset=Know-Show subset2025.12 | 64 | 13.8 | 80 | 52 | |
| Gemini-2.5-FlashTemporal Window=6 seconds, Subset=Know-Show subset2025.12 | 64 | 13.8 | 84 | 56 | |
| Gemini-2.5-ProTemporal Window=2 seconds, Subset=Know-Show subset2025.12 | 60 | 6.6 | 84 | 60 | |
| Gemini-2.5-ProTemporal Window=4 seconds, Subset=Know-Show subset2025.12 | 60 | 6.6 | 88 | 60 | |
| Gemini-2.5-ProTemporal Window=6 seconds, Subset=Know-Show subset2025.12 | 60 | 6.6 | 92 | 60 | |
| Qwen2.5 VL 7BTemporal Window=2 seconds2025.12 | 55.8 | 6.6 | 28.4 | 16.1 | |
| Qwen2.5 VL 7BTemporal Window=4 seconds2025.12 | 55.8 | 6.6 | 48.1 | 29.6 | |
| Qwen2.5 VL 7BTemporal Window=6 seconds2025.12 | 55.8 | 6.6 | 57.8 | 34.1 | |
| Ovis2.5-9BTemporal Window=2 seconds2025.12 | 53.6 | 23.9 | 28.2 | 18.2 | |
| Ovis2.5-9BTemporal Window=4 seconds2025.12 | 53.6 | 23.9 | 44.8 | 28.6 | |
| Ovis2.5-9BTemporal Window=6 seconds2025.12 | 53.6 | 23.9 | 56 | 34 | |
| LLaVA-OneVisionTemporal Window=2 seconds2025.12 | 48.9 | 25.1 | 23.7 | 12.8 | |
| LLaVA-OneVisionTemporal Window=4 seconds2025.12 | 48.9 | 25.1 | 45.7 | 26.9 | |
| LLaVA-OneVisionTemporal Window=6 seconds2025.12 | 48.9 | 25.1 | 54.4 | 31.1 | |
| GRAMTemporal Window=2 seconds, Base Model=Qwen 2.5 7B2025.12 | 48.7 | 22.1 | 36.1 | 23.7 | |
| GRAMTemporal Window=4 seconds, Base Model=Qwen 2.5 7B2025.12 | 48.7 | 22.1 | 49.9 | 30.1 | |
| GRAMTemporal Window=6 seconds, Base Model=Qwen 2.5 7B2025.12 | 48.7 | 22.1 | 61.8 | 34.7 | |
| MiniCPM-V 4.5Temporal Window=2 seconds2025.12 | 47.9 | 58.3 | 30.9 | 19.3 | |
| MiniCPM-V 4.5Temporal Window=4 seconds2025.12 | 47.9 | 58.3 | 46.8 | 28.7 | |
| MiniCPM-V 4.5Temporal Window=6 seconds2025.12 | 47.9 | 58.3 | 57.5 | 34.3 | |
| VideoLLaVATemporal Window=2 seconds2025.12 | 29.8 | 53.3 | 9.6 | 4.9 | |
| VideoLLaVATemporal Window=4 seconds2025.12 | 29.8 | 53.3 | 17.3 | 6.4 | |
| VideoLLaVATemporal Window=6 seconds2025.12 | 29.8 | 53.3 | 22.9 | 8.1 |