CG-Eval

CG-Eval is a test benchmark for Chinese large model generation capabilities jointly developed by Oracle AI Research Institute and LanguageX AI Lab. In this test, the tested Chinese large language model required 11,000 test results for 55 sub-subjects under six major subject categories: technology and engineering, humanities and social sciences, mathematical calculations, medical qualification examinations, judicial examinations, and certified public accountant examinations. Provide accurate and relevant answers to different types of questions. We have designed a composite scoring system. For non-calculation questions, each noun explanation question and short answer question has a standard reference answer. Multiple standards are used for scoring and then weighted and summed. For calculation questions, we will extract the final calculation results and solution process, and then give comprehensive scores.