Vision-Centric Evaluation on CV-Bench 2D
63.8ScoreV-GIFT
Evaluation results
| Method | Links | |
|---|---|---|
| V-GIFTModel=LLaVA-1.5-Qwen2.5-7B, Training Protocol=LoRA2026.04 | 63.8 | |
| LaVerVisual Encoder Model=Qwen-ViT2025.12 | 61.6 | |
| VIRALModel=LLaVA-1.5-Qwen2.5-7B, Training Protocol=LoRA2026.04 | 60.5 | |
| BaselineVisual Encoder Model=Qwen-ViT2025.12 | 59.9 | |
| BaselineModel=LLaVA-1.5-Qwen2.5-7B, Training Protocol=LoRA2026.04 | 59.9 | |
| LaVerVisual Encoder Model=SigLIP22025.12 | 55.6 | |
| BaselineVisual Encoder Model=SigLIP22025.12 | 52.2 | |
| LaVerVisual Encoder Model=AIMv22025.12 | 51.1 | |
| BaselineVisual Encoder Model=AIMv22025.12 | 50.6 | |
| LaVerVisual Encoder Model=CLIP2025.12 | 48.9 | |
| BaselineVisual Encoder Model=CLIP2025.12 | 45.4 | |
| LaVerVisual Encoder Model=DINOv22025.12 | 44.9 | |
| BaselineVisual Encoder Model=DINOv22025.12 | 42.1 | |
| LaVerVisual Encoder Model=MLP + Qwen 2.52025.12 | 38 | |
| BaselineVisual Encoder Model=MLP + Qwen 2.52025.12 | 37.4 |