Visual Reasoning on VisualProbe (VP) cross-domain (test)
0.4357AccuracyVIRC-7B
Evaluation results
| Method | Links | |
|---|---|---|
| VIRC-7B#Params=7B2025.12 | 0.4357 | |
| VIRC-3B#Params=3B2025.12 | 0.3362 | |
| Qwen2.5-VL-7B-Instruct#Params=7B2025.12 | 0.2967 | |
| Hint-GRPO-Qwen2.5-VL-3B#Params=3B2025.12 | 0.2831 | |
| Hint-GRPO-Qwen2-VL-7B#Params=7B2025.12 | 0.2785 | |
| Qwen2.5-VL-3B-Instruct#Params=3B2025.12 | 0.2725 | |
| InternVL2.5-8B#Params=8B2025.12 | 0.2607 | |
| GPT-4o2025.12 | 0.247 | |
| Qwen3-VL-8B-Instructw/ tool#Params=8B2025.12 | 0.2307 | |
| InternVL2.5-8B-MPO#Params=8B2025.12 | 0.2102 | |
| LLaVA-OV-Qwen2-7b-ov#Params=7B2025.12 | 0.207 | |
| R1-VL-7B#Params=7B2025.12 | 0.2026 | |
| MINT-CoT-7B#Params=7B2025.12 | 0.1784 | |
| DeepSeek-VL2#Params=4.5B2025.12 | 0.1712 | |
| MM-Eureka#Params=7B2025.12 | 0.1169 |