3D Question Answering on 3D MM-VET 42 (test)
57.6AccuracyMPM
Evaluation results
| Method | Links | |
|---|---|---|
| MPMTuning=PEFT(78M), Input=3D point cloud2026.05 | 57.6 | |
| Qwen3-Omni-30B-A3BTuning=–, Input=3-view 2D image2026.05 | 56.1 | |
| ShapeLLM-13BTuning=Full(~13B), Input=3D point cloud2026.05 | 53.1 | |
| AdapterTuning=PEFT(9.5M), Input=3D point cloud2026.05 | 49.7 | |
| ShapeLLM-7BTuning=Full(~7B), Input=3D point cloud2026.05 | 47.4 | |
| PointLLM-13BTuning=Full(~13B), Input=3D point cloud2026.05 | 46.6 | |
| PointLLM-7BTuning=Full(~7B), Input=3D point cloud2026.05 | 41.2 | |
| PointBind&LLMTuning=LoRA(~7B), Input=3D point cloud2026.05 | 23.5 |