Path-level reasoning on UNOBench real (Medium)
76.6SR-P (%)UNOGrasp
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| UNOGraspPrompting=Oracle (with SoM), Protocol=Standard2025.11 | 76.6 | 59.6 | 64.4 | 0.4 | |
| UNOGraspPrompting=Natural Language Prompting, Protocol=Standard2025.11 | 62.9 | 40.1 | 47.3 | 0.63 | |
| Qwen2.5-VLPrompting=Oracle (with SoM), Protocol=SFT2025.11 | 62.4 | 41.6 | 48 | — | |
| Qwen2.5-VLPrompting=Natural Language Prompting, Protocol=SFT2025.11 | 61.6 | 37.9 | 45.4 | — | |
| Gemini Robotics-ER 1.5Prompting=Natural Language Prompting, Protocol=Standard2025.11 | 47.3 | 32.7 | 37.1 | 0.89 | |
| Gemini Robotics-ER 1.5Prompting=Natural Language Prompting, Protocol=ICL2025.11 | 44.7 | 34.6 | 36.8 | 0.88 | |
| Gemini Robotics-ER 1.5Prompting=Oracle (with SoM), Protocol=ICL2025.11 | 44.2 | 34.3 | 36.4 | 0.89 | |
| Gemini Robotics-ER 1.5Prompting=Oracle (with SoM), Protocol=Standard2025.11 | 40.8 | 28.2 | 32.1 | 0.9 | |
| Qwen2.5-VLPrompting=Oracle (with SoM), Protocol=ICL2025.11 | 38.3 | 28.2 | 30 | 0.68 | |
| Qwen2.5-VLPrompting=Natural Language Prompting, Protocol=ICL2025.11 | 19.6 | 13.8 | 15.5 | 0.86 |