Multimodal Perception on HR-Bench 4K
84.4AccuracyZwZ
Evaluation results
| Method | Links | |
|---|---|---|
| ZwZModel Size=8B2026.02 | 84.4 | |
| ZwZModel Size=7B2026.02 | 75.4 | |
| AGILEModel Size=7B2026.02 | 73 | |
| SkiLa2025.12 | 72 | |
| LVR 7B*Model Size=7B, Tested by authors=true2025.12 | 69.6 | |
| Qwen2.5-VL 7BModel Size=7B2025.12 | 68.5 | |
| SkiLa-V2025.12 | 68.5 | |
| Direct SFT2025.12 | 65.9 | |
| Vision-R1 7B*Model Size=7B, Tested by authors=true2025.12 | 64.8 | |
| LLaVA-OneVision 7BModel Size=7B2025.12 | 63 | |
| GPT-4o2025.12 | 59 |