Object state change reasoning on ObjChangeVR (All Trajectories)
75.4EM@0.8ObjChangeVR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ObjChangeVRModel=GPT-4o2026.02 | 75.4 | 77 | 77.4 | |
| ObjChangeVRModel=Gemini 2.0 Flash2026.02 | 71.3 | 73.9 | 74.1 | |
| ObjChangeVRModel=GPT-4o mini2026.02 | 65.3 | 65.6 | 65.7 | |
| Viewpoint-RetrievalModel=Gemini 2.0 Flash2026.02 | 62.8 | 64.5 | 64.5 | |
| Image-CLIPModel=Gemini 2.0 Flash2026.02 | 61.1 | 63 | 63 | |
| Viewpoint-RetrievalModel=GPT-4o2026.02 | 60.1 | 63.1 | 63.5 | |
| Image-CLIPModel=GPT-4o2026.02 | 59.2 | 61.9 | 62 | |
| Caption-CLIPModel=GPT-4o2026.02 | 52.9 | 57.4 | 57 | |
| Viewpoint-RetrievalModel=GPT-4o mini2026.02 | 49.4 | 44.9 | 42.9 | |
| Image-CLIPModel=GPT-4o mini2026.02 | 48.9 | 44.1 | 42 | |
| Caption-CLIPModel=GPT-4o mini2026.02 | 46.4 | 42 | 40 | |
| Caption-CLIPModel=Gemini 2.0 Flash2026.02 | 46.2 | 47.1 | 45.5 |