Categorical Visual Question Answering on InstaOrder Depth
63.5AccuracySN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SNBase Model=LLaVA-v1.5-7b, Probing Set Size=3,000, Configuration=Best probed2026.03 | 63.5 | 69.1 | |
| SNBase Model=Qwen-VL-4b-Instruct, Probing Set Size=3,000, Configuration=Best probed2026.03 | 62.8 | 69.1 | |
| Qwen-VL-4b-InstructProbing Set Size=3,0002026.03 | 61.8 | 63.1 | |
| LLaVA-v1.5-7bProbing Set Size=3,0002026.03 | 61.7 | 67.7 |