Error Detection on CoSPlan
9.1Robo-VQA-E ScoreCoT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CoTVLM=Qwen2 VL-8B2025.12 | 9.1 | 20.8 | 30.6 | |
| VanillaVLM=Qwen2 VL-8B2025.12 | 9.2 | 20.5 | 32.3 | |
| SGVLM=Qwen2 VL-8B2025.12 | 9.6 | 20.7 | 35.2 | |
| SGIVLM=Qwen2 VL-8B2025.12 | 10.1 | 21.3 | 35.7 | |
| Random2025.12 | 15.4 | 33.3 | 33.3 | |
| VanillaVLM=Janus-pro-7B2025.12 | 17.5 | 20.5 | 29.3 | |
| CoTVLM=Janus-pro-7B2025.12 | 18.1 | 19.1 | 31 | |
| VanillaVLM=Intern-VLM2025.12 | 24.3 | 32.8 | 36.5 | |
| CoTVLM=Intern-VLM2025.12 | 25.2 | 33.1 | 37.9 | |
| SGVLM=Janus-pro-7B2025.12 | 26.1 | 21 | 27.6 | |
| SGVLM=Intern-VLM2025.12 | 26.1 | 33.4 | 37.3 | |
| SGIVLM=Janus-pro-7B2025.12 | 27.6 | 21.6 | 33.2 | |
| SGIVLM=Intern-VLM2025.12 | 31.5 | 34.8 | 42.9 | |
| VanillaVLM=CoG-VLM2025.12 | 32.1 | 6.4 | 41.3 | |
| CoTVLM=CoG-VLM2025.12 | 33.4 | 8.4 | 43.1 | |
| SGVLM=CoG-VLM2025.12 | 35.3 | 13.3 | 44.5 | |
| SGIVLM=CoG-VLM2025.12 | 38.7 | 11 | 46.1 | |
| SGVLM=GPT-4o2025.12 | 44.2 | 35.3 | 42.1 | |
| CoTVLM=GPT-4o2025.12 | 45.3 | 40.3 | 35.1 | |
| SGIVLM=GPT-4o2025.12 | 57.4 | 41.1 | 50.7 |