Note Generation on MM-HEALTHCN Image
96.5Relevance (R)EVONOTE-FM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| EVONOTE-FMBackbone=MedGemma-27B, Memory=frozen warm-up2026.06 | 96.5 | 85.25 | 78.25 | 86.13 | |
| EVONOTEBackbone=MedGemma-27B2026.06 | 96.5 | 86.75 | 80.5 | 89.25 | |
| Grok-1.52026.06 | 94.5 | 76.25 | 71.75 | 77.88 | |
| ReMemBackbone=MedGemma-27B2026.06 | 94 | 82 | 72 | 72.75 | |
| EVONOTE-NMBackbone=MedGemma-27B, Memory=removed retrieval and update2026.06 | 93 | 77.25 | 71 | 82 | |
| DeepResearch2026.06 | 92.5 | 75.75 | 70 | 76.75 | |
| ExpRAGBackbone=MedGemma-27B2026.06 | 91 | 76.5 | 70.25 | 81.25 | |
| EVONOTE-NM-NABackbone=MedGemma-27B, Memory=removed retrieval and update, Claim Analyzer=removed2026.06 | 90.5 | 74.5 | 69 | 79 | |
| CrowdNotes+Backbone=MedGemma-27B2026.06 | 87.5 | 76.5 | 65.75 | 70.63 | |
| Human Baseline2026.06 | 74.75 | 55.5 | 40 | — | |
| Gemini-1.5-Flash2026.06 | 63.5 | 57.25 | 54 | 75.88 | |
| GPT-4o2026.06 | 36.25 | 28.25 | 27 | 63.25 |