Domain-Specific Visual Question Answering on RSVQA
89.9AccuracyMaLoRA
Evaluation results
| Method | Links | |
|---|---|---|
| MaLoRAModel=LLaVA-1.5-7B, Training examples=16.1k2025.10 | 89.9 | |
| LoRAModel=LLaVA-1.5-7B, Training examples=16.1k2025.10 | 89.1 | |
| QLoRAModel=LLaVA-1.5-7B, Training examples=16.1k2025.10 | 88.4 | |
| MaLoRAModel=Qwen3-VL-8B, Training examples=16.1k2025.10 | 88.3 | |
| MaLoRAModel=Qwen2.5-VL-7B, Training examples=16.1k2025.10 | 87.9 | |
| LoRAModel=Qwen2.5-VL-7B, Training examples=16.1k2025.10 | 87.7 | |
| LoRAModel=Qwen3-VL-8B, Training examples=16.1k2025.10 | 86.8 | |
| QLoRAModel=Qwen2.5-VL-7B, Training examples=16.1k2025.10 | 63.5 | |
| BaseModel=Qwen2.5-VL-7B, Training examples=16.1k2025.10 | 62.3 | |
| BaseModel=Qwen3-VL-8B, Training examples=16.1k2025.10 | 61 | |
| QLoRAModel=Qwen3-VL-8B, Training examples=16.1k2025.10 | 58.9 | |
| BaseModel=LLaVA-1.5-7B, Training examples=16.1k2025.10 | 50.9 |