Pathology Visual Question Answering on PathMMU (val)
62.4Overall AccuracyScaleReasoner-R1
Evaluation Results
| Method | Links | |
|---|---|---|
| ScaleReasoner-R1Category=Pathological VLM, Backbone=PathoR1-7B2026.06 | 62.4 | |
| Patho-R1Category=Pathological VLM, Backbone=PathoR1-7B2026.06 | 62.3 | |
| CLOVERCategory=Pathological VLM2026.06 | 53.1 | |
| Lingshu-7BCategory=Medical VLM, Parameters=7B2026.06 | 51.9 | |
| GPT-4VCategory=General domain VLM, Source=PathMMU paper2026.06 | 49.3 | |
| SFT + RLCategory=Ablation Study, Backbone=PathoR1-7B2026.06 | 48.3 | |
| SFTCategory=Ablation Study, Backbone=PathoR1-7B2026.06 | 47.8 | |
| Qwen2.5-VL-7BCategory=General domain VLM, Parameters=7B2026.06 | 44.3 | |
| HuatuoGPT-V-7BCategory=Medical VLM, Parameters=7B2026.06 | 43.2 | |
| Gemini Pro VisionCategory=General domain VLM, Source=PathMMU paper2026.06 | 41.9 | |
| Quilt-LLaVACategory=Pathological VLM2026.06 | 33.8 | |
| Frequent ChoiceCategory=Baseline, Source=PathMMU paper2026.06 | 27.5 | |
| Random ChoiceCategory=Baseline, Source=PathMMU paper2026.06 | 24.6 | |
| LLaVA-Med-7BCategory=Medical VLM, Parameters=7B2026.06 | 17.5 |