Alibaba Claims Qwen2-Math Outperforms Rivals in Mathematical Reasoning
Alibaba has announced that its new open-source large language model, Qwen2-Math, achieves state-of-the-art performance in mathematical reasoning. The tech giant claims the model handles algebra and geometry problems with an accuracy rate of 84%, surpassing prominent competitors such as OpenAI’s GPT-4o and Google’s Gemini 1.5 Pro. Built upon Alibaba’s foundational Qwen2 model, Qwen2-Math was trained on a extensive, high-quality corpus of mathematical data. Currently, the model supports only English, but a bilingual version is scheduled for imminent release. To validate its capabilities, the Alibaba team subjected the model to rigorous benchmarks, including complex questions from China’s national college entrance examination (GaoKao) and the American Invitational Mathematics Examination (AIME). The results reportedly demonstrated the model's proficiency in solving advanced mathematical challenges. This development highlights Alibaba's continued investment in specialized artificial intelligence applications, aiming to compete directly with global leaders in the AI sector by offering superior performance in specific technical domains like mathematics.
Wire timeline
Alibaba Claims Qwen2-Math Outperforms Rivals in Mathematical Reasoning
Alibaba has announced that its new open-source large language model, Qwen2-Math, achieves state-of-the-art performance in mathematical reasoning. The tech giant claims the model handles algebra and geometry problems with an accuracy rate of 84%, surpassing prominent competitors such as OpenAI’s GPT-4o and Google’s Gemini 1.5 Pro. Built upon Alibaba’s foundational Qwen2 model, Qwen2-Math was trained on a extensive, high-quality corpus of mathematical data. Currently, the model supports only English, but a bilingual version is scheduled for imminent release. To validate its capabilities, the Alibaba team subjected the model to rigorous benchmarks, including complex questions from China’s national college entrance examination (GaoKao) and the American Invitational Mathematics Examination (AIME). The results reportedly demonstrated the model's proficiency in solving advanced mathematical challenges. This development highlights Alibaba's continued investment in specialized artificial intelligence applications, aiming to compete directly with global leaders in the AI sector by offering superior performance in specific technical domains like mathematics.
TechNode