Data Science and Technology Towards AGI Part I: Tiered Data Management

Benchmark Model Rank Results
instruction-following-on-multi-ifMiniCPM5-2B#6Accuracy: 71.8
social-commonsense-reasoning-on-siqaMiniCPM-1.2B (Math-L3)#8Accuracy: 43.35
mathematical-reasoning-on-hmmt-feb-2026MiniCPM5-2B#9Accuracy: 63.8
multiple-choice-question-answering-mcqa-on-supergpqaMiniCPM5-2B#18Accuracy: 40.8
instruction-following-on-ifbenchMiniCPM5-2B#25Accuracy: 66.3
multi-task-language-understanding-on-mmlu-reduxMiniCPM5-2B#26Accuracy: 84.7
mathematical-reasoning-on-aime-2025MiniCPM5-2B#30Accuracy: 86.5
multi-task-language-understanding-on-mmlu-proMiniCPM5-2B#71Accuracy: 70.8
common-sense-reasoning-on-arc-easyMiniCPM-1.2B (Math-L3)#78Accuracy: 59.79
common-sense-reasoning-on-winograndeMiniCPM-1.2B (Math-L3)#125Accuracy: 59.04
mathematical-reasoning-on-math-500MiniCPM-1.2B (Math-L3)#129Accuracy: 37.02
sentence-completion-on-hellaswagMiniCPM-1.2B (Math-L3)#159Accuracy: 58.27