Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Explanation Quality Evaluation on PubMedQA (test)

39.7Reasoning Soundness Loss

PubMed Reasoner

39.10443.12747.1551.173Mar 28, 2026
Updated 2mo ago

Evaluation Results

MethodLinks
39.75.754.63.58440.83.855.43.60139.27.453.43.58438.75.455.93.587
2026.03
54.65.739.73.41655.43.840.83.42153.47.439.23.43855.95.438.73.424