doximity/bedside-bench

BedsideBench BedsideBench is a clinical question-answering benchmark with 500 cases and public grading rubrics. It covers factual questions, calculations, false-premise queries, and free-text management plans. Each of the 10 benchmarks contains 50 cases. The public train split (250 cases) is in this repository and on Hugging Face. The held-out test split is not published. Not for clinical use. BedsideBench evaluates AI systems. It is not medical advice and must not be used for… See the full description on the dataset page: https://huggingface.co/datasets/doximity/bedside-bench.

<iframe src="https://silytics.si/embed/item/b4481fc0-c318-400c-a8c3-cc8b91466b11" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.