Residential Floor Plan Furnishing on CubiCasa5K Bedroom (test)
68Win Rate (Ours)Architect-Ant
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Architect-AntOpponent Model=GLM-5V-Turbo, Judge (VLM-as-judge)=Present, Sample size (n)=1002026.06 | 68 | 29 | 3 | |
| Architect-AntOpponent Model=Kimi K2.5, Judge (VLM-as-judge)=Present, Sample size (n)=1002026.06 | 48 | 51 | 1 |