Manipulating Point Prediction on Manipulating Point Prediction (Seen)
31.4Success RateGRPO-MA
Evaluation Results
| Method | Links | |
|---|---|---|
| GRPO-MATN=4, AN=42025.09 | 31.4 | |
| CoT-SFT2025.09 | 28.18 | |
| ManipLLM-7B2025.09 | 22.8 | |
| GRPOTN=4, AN=12025.09 | 10.75 | |
| GRPO-NoThinkTN=0, AN=162025.09 | 10.6 | |
| SFT2025.09 | 9.17 | |
| Qwen2.5-VL-3B2025.09 | 4.73 |