Diagnostic Reasoning on DiReCT (test)
66.5Reasoning RecallMultiDx
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MultiDxModel Category=Agentic Method, Backbone=DeepSeek-R1, Runs=Averaged over 3 random runs2026.04 | 66.5 | 33.3 | 50.3 | 58.7 | |
| Self-refinementModel Category=Agentic Method, Backbone=DeepSeek-R12026.04 | 66.2 | 30 | 46.6 | 58.6 | |
| OpenAI-DRModel Category=Agentic Method, Backbone=DeepSeek-R12026.04 | 58.6 | 29.7 | 45.2 | 47.9 | |
| DeepSeek-R1Model Category=Base Model2026.04 | 47.3 | 29.3 | 41.3 | 47.3 |