Classification on CT-Rate (test)
57.89Micro PrecisionAstra
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AstraBootstrap iterations=2,000, Confidence Interval=95%2026.05 | 57.89 | 52.42 | 55.02 | |
| LLaVABootstrap iterations=2,000, Confidence Interval=95%2026.05 | 54.06 | 27.51 | 36.47 | |
| R2GenGPTBootstrap iterations=2,000, Confidence Interval=95%2026.05 | 51.18 | 35.8 | 42.13 | |
| HuluMed-32BModel size=32B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 46.87 | 32.1 | 38.1 | |
| HuluMed-7BModel size=7B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 46.82 | 25.15 | 32.72 | |
| HuluMed-14BModel size=14B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 43.73 | 32.12 | 37.04 | |
| Gemini-3Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 41.34 | 18.28 | 25.35 | |
| MedGemma-4BModel size=4B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 36.43 | 1.01 | 1.97 | |
| Qwen3-VL-8BModel size=8B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 35.61 | 6.95 | 11.63 | |
| MedGemma-27BModel size=27B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 29.07 | 0.5 | 0.98 | |
| RadFMBootstrap iterations=2,000, Confidence Interval=95%2026.05 | 28.5 | 6.44 | 10.5 | |
| Lingshu-7BModel size=7B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 24.62 | 4.49 | 7.59 | |
| M3DBootstrap iterations=2,000, Confidence Interval=95%2026.05 | 24.46 | 9.97 | 14.17 | |
| Lingshu-32BModel size=32B, Bootstrap iterations=2,000, Confidence Interval=95%2026.05 | 21.02 | 5.73 | 9.01 |