Two-choice selection on Science-T2I Simple
94.92Physics AccuracySciScore
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SciScore2025.04 | 94.92 | 80.95 | 100 | 93.14 | |
| Human Eval2025.04 | 87.67 | 75.85 | 95.29 | 87.01 | |
| LLaVA-OV-7B2025.04 | 68.22 | 57.82 | 64.71 | 65.05 | |
| InternVL2.5-8B2025.04 | 67.8 | 62.24 | 84.41 | 70.79 | |
| GPT-4o miniChain-of-Thought (CoT)=true2025.04 | 67.04 | 76.87 | 90 | 74.97 | |
| GPT-4o miniChain-of-Thought (CoT)=false2025.04 | 61.97 | 73.81 | 86.76 | 70.83 | |
| Qwen2-VL-7B2025.04 | 60.03 | 67.01 | 68.82 | 63.79 | |
| SigLIP ViT-SO-142025.04 | 59.63 | 53.17 | 55.94 | 57.23 | |
| CLIP-H2025.04 | 55.08 | 52.38 | 55.88 | 54.69 | |
| BLIPScore2025.04 | 50.35 | 43.08 | 59.86 | 55 | |
| Random Guess2025.04 | 50 | 50 | 50 | 50 |