Sentence-level Confidence Calibration on MIMIC-CXR (test)
0.283GREENMedGemma Base
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MedGemma Base2026.03 | 0.283 | — | — | |
| Verbalize Base2026.03 | 0.268 | 0.438 | 0.558 | |
| Verbalize Supervised2026.03 | 0.267 | 0.446 | 0.552 | |
| ConRad2026.03 | 0.262 | 0.239 | 0.633 |