Visual Question Answering on CXR-VQA (test)
92.68PresenceDeepMedix-R1
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| DeepMedix-R12025.09 | 92.68 | 64.47 | 33.01 | 98 | 48.27 | 23.04 | 65.89 | |
| DeepMedix2025.09 | 86.89 | 60.28 | 32.15 | 96.17 | 39.7 | 18.34 | 61.41 | |
| CheXagent2025.09 | 78.77 | 54.53 | 31.51 | 77 | 6.43 | 5.37 | 49.72 | |
| MedGemma2025.09 | 65.98 | 58.55 | 30.3 | 73.67 | 7.91 | 8.72 | 47.31 | |
| HuatuoGPT-Vision2025.09 | 65.25 | 34.36 | 26.05 | 70.17 | 1.48 | 4.7 | 38.81 | |
| Qwen2.5-VL-7Bparameters=7B, mode=think2025.09 | 58.85 | 21.97 | 22.03 | 62 | 1.32 | 1.57 | 32.13 | |
| Qwen2.5-VL-32Bparameters=32B2025.09 | 54.51 | 29.66 | 17.2 | 69.5 | 4.45 | 3.8 | 33.86 | |
| Qwen2.5-VL-7Bparameters=7B2025.09 | 22.62 | 41.28 | 28.14 | 73 | 0.33 | 6.49 | 30.07 | |
| DeepSeek-VL-7Bparameters=7B2025.09 | 14.18 | 10.51 | 14.95 | 15 | 2.97 | 4.25 | 11.06 | |
| LLaVA-Rad2025.09 | 13.93 | 8.95 | 3.55 | 0 | 1.64 | 0.44 | 6.62 |