Vision-Language Compositionality on SugarCrepe
88.06AccuracyDINOv2
Evaluation Results
| Method | Links | |
|---|---|---|
| DINOv2LLM=Qwen2-7B2026.02 | 88.06 | |
| mFRESALLM=Qwen2-7B2026.02 | 83.17 | |
| DINOv2LLM=Phi22026.02 | 82.05 | |
| FT-DINOSAURLLM=Qwen2-7B2026.02 | 81.24 | |
| VQDINOMLPLLM=Qwen2-7B2026.02 | 80.16 | |
| StableLSDLLM=Qwen2-7B2026.02 | 78.98 | |
| DINOSAURv2LLM=Qwen2-7B2026.02 | 78.18 | |
| mFRESALLM=Phi22026.02 | 77.27 | |
| DINOSAURv2LLM=Phi22026.02 | 75.98 | |
| Slot DiffusionLLM=Qwen2-7B2026.02 | 74.53 | |
| SPOTLLM=Qwen2-7B2026.02 | 74.08 | |
| VQDINOMLPLLM=Phi22026.02 | 73.05 | |
| StableLSDLLM=Phi22026.02 | 72.92 | |
| DINOSAURLLM=Qwen2-7B2026.02 | 72.45 | |
| SPOTLLM=Phi22026.02 | 71.65 | |
| FT-DINOSAURLLM=Phi22026.02 | 70.94 | |
| Slot DiffusionLLM=Phi22026.02 | 70.39 | |
| DINOSAURLLM=Phi22026.02 | 67.85 | |
| Blind-VLMLLM=Qwen2-7B2026.02 | 53.29 | |
| Blind-VLMLLM=Phi22026.02 | 49.42 |