Multi-image Medical Visual Question Answering on Med-MIM Held-in
74.86Temporal (C)Med-Mantis
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Med-Mantistraining_dataset=Med-MIM instruction dataset2025.05 | 74.86 | 46.87 | 29.74 | 42.05 | 52.88 | 40.33 | 80.4 | 49.2 | |
| MIM-LLAVA-Medtraining_dataset=Med-MIM instruction dataset2025.05 | 74.31 | 44.5 | 22.91 | 41.74 | 50 | 40.36 | 46.19 | 43.09 | |
| Mantis-8B2025.05 | 68.89 | 37.32 | 15.2 | 24.05 | 38.94 | 31.47 | 35.71 | 22.72 | |
| GPT-4o2025.05 | 47.95 | 37.28 | 16.96 | 30.18 | 29.81 | 37.2 | 42.3 | 33.81 | |
| InternVL2-8B2025.05 | 46.07 | 35.85 | 16.08 | 32.12 | 34.13 | 36.09 | 41.05 | 33.92 | |
| Med-Flamingo-9B2025.05 | 39.65 | 24.11 | 15.64 | 22.01 | 32.21 | 27.03 | 28.72 | 27.2 | |
| LLaVA-Med-7B2025.05 | 38.87 | 37.41 | 14.32 | 34.9 | 33.17 | 36.65 | 41.13 | 33.83 | |
| LLaVA-Med†setup=without composed instruction dataset2025.05 | 35.44 | 7.09 | 8.59 | 6.84 | 16.83 | 3.71 | 5.06 | 9.26 | |
| deepseek-VL-7B2025.05 | 29.9 | 36.07 | 14.54 | 34.47 | 32.69 | 34.34 | 29.01 | 31.08 | |
| Mantis†setup=without composed instruction dataset2025.05 | 5.97 | 9.55 | 14.54 | 18 | 13.94 | 8.86 | 11.69 | 26.48 |