Error Detection on CoSPlan Blocks-World-E
44.5AccuracyCoG-VLM
Evaluation Results
| Method | Links | |
|---|---|---|
| CoG-VLMReasoning Strategy=Scene Graph2025.12 | 44.5 | |
| CoG-VLMReasoning Strategy=Chain-of-Thought2025.12 | 43.1 | |
| GPT-4oReasoning Strategy=Scene Graph2025.12 | 42.1 | |
| CoG-VLMReasoning Strategy=Vanilla2025.12 | 41.3 | |
| Intern-VLMReasoning Strategy=Chain-of-Thought2025.12 | 37.9 | |
| Intern-VLMReasoning Strategy=Scene Graph2025.12 | 37.3 | |
| Intern-VLMReasoning Strategy=Vanilla2025.12 | 36.5 | |
| Qwen2 VL-8BReasoning Strategy=Scene Graph2025.12 | 35.2 | |
| GPT-4oReasoning Strategy=Chain-of-Thought2025.12 | 35.1 | |
| Qwen2 VL-8BReasoning Strategy=Vanilla2025.12 | 32.3 | |
| Janus-pro-7BReasoning Strategy=Chain-of-Thought2025.12 | 31 | |
| Qwen2 VL-8BReasoning Strategy=Chain-of-Thought2025.12 | 30.6 | |
| Janus-pro-7BReasoning Strategy=Vanilla2025.12 | 29.3 | |
| Janus-pro-7BReasoning Strategy=Scene Graph2025.12 | 27.6 | |
| Random2025.12 | 26.1 |