Multimodal Explanation on VQA-X
94.11F1 ScoreLLaVA
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LLaVAfine-tuned=true2025.12 | 94.11 | 0.07 | 0.022 | 0.011 | 0.008 | 0.679 | |
| PaLiGemmafine-tuned=true2025.12 | 94.08 | 0.112 | 0.034 | 0.026 | 0.024 | 0.712 | |
| LLaVAshots=2, fine-tuned=false2025.12 | 91.11 | 0.033 | 0.014 | 0.007 | 0.004 | 0.653 | |
| LLaVAshots=0, fine-tuned=false2025.12 | 88.12 | 0.045 | 0.022 | 0.011 | 0.002 | 0.642 | |
| PaLiGemmashots=2, fine-tuned=false2025.12 | 35.74 | 0.004 | 0.001 | 0.001 | 0 | 0.651 | |
| PaLiGemmashots=0, fine-tuned=false2025.12 | 35.05 | 0.004 | 0.001 | 0.001 | 0 | 0.653 |