You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Does medical-reasoning fine-tuning make a small LLM safer or confidently wrong? A controlled before/after calibration study on Qwen3-4B + MedGemma-4B (single free Kaggle T4). Finding: internal cali…
Clinical tools for DeepSeek Harness: handover and discharge drafts, de-identification, completeness checks, and cited lookups. Draft-only, not a medical device.