待验证60% 置信基准精确时间
MMLU涵盖57个学科、近16,000道选择题；HumanEval由OpenAI开发包含164个编程题；GSM8K包含8,500道小学数学应用题
2
来源数
60%
置信度
长期有效
时效性
2026/8/31
首次发现
来源
涉及实体
相关事实
待验证HumanEval由164道Python编程题组成，GSM8K包含8,500道小学数学应用题75% 相似已验证MMLU contains approximately 15,000 multiple-choice questions across 57 subjects, covering knowledge from high school to graduate level74% 相似已验证MMLU（Massive Multitask Language Understanding）涵盖57个学科的选择题，从人文社科到STEM测试模型的知识广度与深度73% 相似待验证MMLU由加州大学伯克利分校的Dan Hendrycks等人于2020年提出，涵盖57个学科领域、约15,000道选择题69% 相似已验证GSM8K, released by OpenAI, contains approximately 8,500 elementary school math word problems designed to test multi-step reasoning ability64% 相似
引用此条事实
Stable URI
https://kongchang.com/claim/829753API
curl https://kongchang.com/api/v1/knowledge/claims/829753MCP
get_claim(id=829753)