Two-choice selection on Science-T2I Complex
86.89Accuracy (Physics)SciScore
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SciScore2025.04 | 86.89 | 91.11 | 100 | 91.19 | |
| Human Eval2025.04 | 84.71 | 85.4 | 89.14 | 86.02 | |
| LLaVA-OV-7B2025.04 | 74.59 | 50 | 76.67 | 70.26 | |
| InternVL2.5-8B2025.04 | 73.77 | 65.56 | 85.83 | 75.33 | |
| GPT-4o miniChain-of-Thought (CoT)=true2025.04 | 72.44 | 70 | 92.5 | 77.16 | |
| GPT-4o miniChain-of-Thought (CoT)=false2025.04 | 69.29 | 70 | 90 | 74.78 | |
| Qwen2-VL-7B2025.04 | 66.8 | 50 | 90.83 | 69.82 | |
| SigLIP ViT-SO-142025.04 | 61.48 | 51.11 | 70 | 61.67 | |
| CLIP-H2025.04 | 56.56 | 44.44 | 76.67 | 59.47 | |
| Random Guess2025.04 | 50 | 50 | 50 | 50 | |
| BLIPScore2025.04 | 49.78 | 60 | 58.33 | 51.54 |