Note Generation on MM-HEALTHCN Text
97.5Relevance (R)EVONOTE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| EVONOTEBackbone=MedGemma-27B2026.06 | 97.5 | 89.25 | 78.75 | 89.38 | |
| EVONOTE-FMBackbone=MedGemma-27B, Memory=frozen warm-up2026.06 | 97.25 | 80.25 | 70.25 | 85.38 | |
| DeepResearch2026.06 | 95.5 | 81.25 | 73.75 | 81.13 | |
| EVONOTE-NMBackbone=MedGemma-27B, Memory=removed retrieval and update2026.06 | 95.5 | 80.5 | 68.5 | 85.25 | |
| Grok-1.52026.06 | 94.75 | 79.5 | 69 | 76.5 | |
| EVONOTE-NM-NABackbone=MedGemma-27B, Memory=removed retrieval and update, Claim Analyzer=removed2026.06 | 90.75 | 74.25 | 64.5 | 81.63 | |
| ReMemBackbone=MedGemma-27B2026.06 | 90.25 | 77.5 | 67 | 76.75 | |
| CrowdNotes+Backbone=MedGemma-27B2026.06 | 88.5 | 83.5 | 68.5 | 69.25 | |
| ExpRAGBackbone=MedGemma-27B2026.06 | 83.25 | 66.75 | 59 | 76.88 | |
| Human Baseline2026.06 | 73 | 53.5 | 35.75 | — | |
| Gemini-1.5-Flash2026.06 | 63.25 | 57.75 | 50.75 | 76.13 | |
| GPT-4o2026.06 | 39.75 | 31.5 | 28.5 | 69.5 |