Visual Question Answering on TextVQA (Accuracy)
58.3AccuracyLLaVA1.5-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA1.5-7B2026.01 | 58.3 | |
| WiSE-FT2026.01 | 57.8 | |
| CARPE-MoE2026.01 | 57.4 | |
| LLaVA1.5-7BFine-tuned=ImageNet2026.01 | 57 | |
| CARPE2026.01 | 55.8 | |
| SPHINX-13B2026.01 | 51.6 | |
| LEVI2026.01 | 49.2 |