Instance Grouping on ScanNet (val)
13.56T-SRQwen3-VL-2B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3-VL-2BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 13.56 | 25.5 | |
| Qwen3-VL-4BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 12.7 | 24.62 | |
| Qwen3-VL-8BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 12.41 | 24.27 | |
| InternVL3-8BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 11.22 | 22.6 | |
| InternVL3.5-8BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 10.83 | 22.02 | |
| Qwen2.5-VL-7BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 10.79 | 22.21 | |
| InternVL3-1BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 10.45 | 20.79 | |
| InternVL3.5-4BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 10.18 | 21.74 | |
| Qwen2.5-VL-3BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 10.12 | 21.53 | |
| InternVL3-2BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Vision-Language Models2026.05 | 10.02 | 21.25 | |
| WAN2.1-T2V-14BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 8.5 | 18.98 | |
| WAN2.1-I2V-14BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 7.31 | 16.83 | |
| WAN2.1-T2V-1.3BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 6.35 | 16.37 | |
| OpenSora2.0Probing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 4.3 | 13.07 | |
| AetherProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 2.39 | 10.54 | |
| CogVideoX-T2V-5BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 2.22 | 10.69 | |
| CogVideoX-T2V-2BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 1.97 | 9.87 | |
| CogVideoX-I2V-5BProbing=Frozen representations, Evaluation=Unified probe, Model Category=Video-Generation Models2026.05 | 1.76 | 9.58 |