{"rank":null,"reference":false,"run_id":"d3265cd5bd82","task":"clinicalqa@0","model":{"id":"gemma-4-26b-a4b","slug":"kgpu/gemma-4-26b-a4b","name":"Gemma 4 26B-A4B","type":"llm","developer":"Google"},"tier":"baseline","metric":{"name":"accuracy","value":0.8736842105263158,"ci_low":0.8535885167464115,"ci_high":0.8937799043062201},"n":1045,"parse_rate":0.999043062200957,"cost_usd":0.20353827000000002,"eval_date":"2026-10-29","url":"https://medbench.kr/r/d3265cd5bd82","status":"done","listed":true,"current_spec":true,"metrics":{"accuracy":{"value":0.8736842105263158,"ci_low":0.8535885167464115,"ci_high":0.8937799043062201,"n":1045,"n_unparsed":1},"parse_rate":0.999043062200957,"error_rate":0.0,"by_category":{"Allergy and Immunology":{"value":0.9166666666666666,"n":36},"Cardiology":{"value":0.9438202247191011,"n":89},"Dermatology":{"value":0.75,"n":12},"Emergency Medicine":{"value":0.96875,"n":32},"Endocrinology":{"value":0.8970588235294118,"n":68},"Endocrinology, metabolic, and nutritional disorders":{"value":1.0,"n":1},"Forensic Medicine":{"value":1.0,"n":1},"Gastroenterology":{"value":0.8839285714285714,"n":112},"General Surgery":{"value":0.8620689655172413,"n":58},"Geriatric Medicine":{"value":1.0,"n":1},"Gynecology":{"value":0.7714285714285715,"n":70},"Hematology and Oncology":{"value":0.9583333333333334,"n":48},"Infectious Diseases":{"value":0.8461538461538461,"n":52},"Musculoskeletal (e.g., costochondritis)":{"value":1.0,"n":1},"Nephrology":{"value":0.8198198198198198,"n":111},"Neurology / Neurosurgery":{"value":0.8979591836734694,"n":49},"Obstetrics":{"value":0.6666666666666666,"n":18},"Ophthalmology":{"value":1.0,"n":1},"Orthopedic Surgery":{"value":0.9166666666666666,"n":24},"Otolaryngology":{"value":1.0,"n":14},"Otolaryngology (ENT)":{"value":0.9285714285714286,"n":14},"Pediatrics":{"value":0.8392857142857143,"n":56},"Pharmacology":{"value":1.0,"n":1},"Psychiatry":{"value":0.8222222222222222,"n":45},"Pulmonology (Respiratory Medicine)":{"value":0.8769230769230769,"n":65},"Radiology":{"value":1.0,"n":1},"Rheumatology":{"value":0.8979591836734694,"n":49},"Urology":{"value":0.9230769230769231,"n":13},"Vascular Surgery":{"value":1.0,"n":2},"nan":{"value":0.0,"n":1}}},"config":{"max_tokens":16384,"prompt":"clinicalqa-p0","temperature":0},"run_key":"e7b81e34d23cf3b63946566261aece9266e6f8b06827c5d400eb1c21c16821bd","harness_version":"p0.1","task_spec_hash":"77e37cd0ca6bded7fba03792ec1922342b3f6a7dc3ce602881ce196e6aa7377c","model_id":"gemma-4-26b-a4b","tokens_in":408483,"tokens_out":542513,"dataset":{"slug":"clinicalqa","title":"ClinicalQA: Korean Medical Knowledge QA Benchmark","license":"CC BY 4.0 (upstream Apache-2.0)","license_url":"https://creativecommons.org/licenses/by/4.0/","citation":"SNUH-HARI. ClinicalQA: Korean Medical Knowledge QA Benchmark. Seoul National University Hospital Healthcare AI Research Institute, 2025. doi:10.23421/2026.0800"},"predictions_url":"https://medbench.kr/artifacts/d3265cd5bd82/predictions.parquet","bibtex":"@misc{medbench_d3265cd5bd82,\n  title = {Gemma 4 26B-A4B on Korean clinical knowledge (KMLE-level MCQ) (clinicalqa@0)},\n  author = {{medbench.kr}},\n  year = {2026},\n  howpublished = {\\url{https://medbench.kr/r/d3265cd5bd82}},\n  note = {Evaluated 2026-10-29, harness p0.1, run key e7b81e34d23c}\n}"}