Alibaba says its math

Alibaba has claimed that its open-source large model Qwen2-Math delivers “state-of-the-art” math competency, saying it handles mathematical problems in algebra and geometry with 84% accuracy, and has outperformed OpenAI’s GPT 4o and Google’s Gemini 1.5 Pro. The math-centered model, which was trained on Alibaba’s foundation Qwen2 model by feeding it a large-scale high-quality math corpus, is currently only supported in English, with a bilingual version coming soon, according to the tech giant. The Alibaba team has challenged Qwen2-Math with multiple benchmarks including China’s latest GaoKao (college entrance exam) math problems and questions from US math competition AIME, which all proved the model’s proficiency in dealing with advanced mathematical problems. [QbitAI, in Chinese]