General Multimodal Performance on POPE, HallusionBench, MMStar Average
69.3Overall ScorePStar
Evaluation Results
| Method | Links | |
|---|---|---|
| PStarBackbone=Qwen2.5-VL-7B2026.05 | 69.3 | |
| PStarBackbone=Qwen2-VL-7B2026.05 | 69 | |
| Qwen2.5-VL-7BModel=Qwen2.5-VL-7B*2026.05 | 67.3 | |
| Qwen2-VL-7BModel=Qwen2-VL-7B*2026.05 | 66.5 | |
| InternVL2-8BModel=InternVL2-8B*2026.05 | 63.6 | |
| PStarBackbone=Qwen2-VL-2B2026.05 | 60.8 | |
| GPT-4V (0409)Model=GPT-4V (0409)*2026.05 | 60.6 | |
| Qwen2-VL-2BModel=Qwen2-VL-2B*2026.05 | 59.5 | |
| LLaMA-3.2-11B-VisionModel=LLaMA-3.2-11B-Vision*2026.05 | 59.4 | |
| InternVL2-2BModel=InternVL2-2B*2026.05 | 57.7 | |
| LLaVA-v1.5-7BModel=LLaVA-v1.5-7B*2026.05 | 48.9 |