Step Completion on CoSPlan Robo-VQA-E
52.2AccuracyGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4oReasoning Strategy=Scene Graph2025.12 | 52.2 | |
| GPT-4oReasoning Strategy=Chain-of-Thought2025.12 | 48.2 | |
| Intern-VLMReasoning Strategy=Scene Graph2025.12 | 25.1 | |
| Intern-VLMReasoning Strategy=Chain-of-Thought2025.12 | 23.5 | |
| Intern-VLMReasoning Strategy=Vanilla2025.12 | 22.1 | |
| CoG-VLMReasoning Strategy=Scene Graph2025.12 | 21.5 | |
| Janus-pro-7BReasoning Strategy=Scene Graph2025.12 | 21.3 | |
| Random2025.12 | 20 | |
| Qwen2 VL-8BReasoning Strategy=Scene Graph2025.12 | 18.9 | |
| Qwen2 VL-8BReasoning Strategy=Chain-of-Thought2025.12 | 17.6 | |
| Qwen2 VL-8BReasoning Strategy=Vanilla2025.12 | 17.1 | |
| Janus-pro-7BReasoning Strategy=Chain-of-Thought2025.12 | 14.7 | |
| Janus-pro-7BReasoning Strategy=Vanilla2025.12 | 14.1 | |
| CoG-VLMReasoning Strategy=Vanilla2025.12 | 13.1 | |
| CoG-VLMReasoning Strategy=Chain-of-Thought2025.12 | 12.5 |