Vision-Language Alignment on VL-Checklist VG-spatial
63.5AccuracyDCSM open-vocab
Evaluation Results
| Method | Links | |
|---|---|---|
| DCSM open-vocabMode=Open-vocabulary2025.03 | 63.5 | |
| CLIP ft.synthTraining=Fine-tuned on synthetic data2025.03 | 56.9 | |
| CLIP ViT-B/16Backbone=ViT-B/162025.03 | 50.8 | |
| DCSM synthTraining=Synthetic data2025.03 | 50.6 |