Diagnosis on DermoBench In-Distribution
90.3Accuracy (4-class)DermoGPT-RL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DermoGPT-RLParams=8B2026.01 | 90.3 | 64.6 | 79.12 | 78.01 | |
| DermoGPT-SFT + CCTParams=8B2026.01 | 89.75 | 64.45 | 78.06 | 77.42 | |
| DermoGPT-RL + CCTParams=8B2026.01 | 89.6 | 65.4 | 79.12 | 78.04 | |
| DermoGPT-SFTParams=8B2026.01 | 89.55 | 64.3 | 77.91 | 77.25 | |
| Human PerformanceParams=-2026.01 | 85 | 77 | 87.54 | 83.18 | |
| Gemini-2.5-FlashParams=/2026.01 | 72.6 | 47.2 | 70.31 | 63.37 | |
| Qwen3-VL-8B (Base)Params=8B2026.01 | 67.2 | 45.35 | 44.77 | 52.44 | |
| QVQ-72B-PreviewParams=72B2026.01 | 64.65 | 47.3 | 57.25 | 56.4 | |
| Qwen3-VL-32BParams=32B2026.01 | 64.25 | 38.05 | 64.08 | 55.46 | |
| GLM-4.5VParams=106B2026.01 | 63.65 | 28.85 | 52.39 | 48.3 | |
| Qwen2.5-VL-72BParams=72B2026.01 | 61.5 | 35.95 | 53.93 | 50.46 | |
| GPT-4o-miniParams=/2026.01 | 59.5 | 34.75 | 65.9 | 53.38 | |
| SkinVL-PubMMParams=7B2026.01 | 57.15 | 38.75 | 52.19 | 49.36 | |
| Claude-Sonnet-4.5-ThinkingParams=/2026.01 | 55.35 | 34.15 | 63.4 | 50.97 | |
| Lingshu-32BParams=32B2026.01 | 53.45 | 38.4 | 49.11 | 46.99 | |
| HuatuoGPT-Vis-7BParams=7B2026.01 | 51.6 | 26.05 | 46.1 | 41.25 | |
| LLaVA-Med-v1.5Params=7B2026.01 | 49.65 | 32.4 | 43.29 | 41.78 | |
| Lingshu-7BParams=7B2026.01 | 49.55 | 31.9 | 43.43 | 41.64 | |
| Llama-3.2-90BParams=90B2026.01 | 47.85 | 51.65 | 51.2 | 50.23 | |
| Nemotron-NanoParams=12B2026.01 | 47.25 | 25.6 | 40.17 | 37.67 | |
| Llama-3.2-11BParams=11B2026.01 | 29.25 | 16.5 | 35.98 | 27.58 |