In order to deal with difficult mathematical problems, Alibaba Cloud’s Queen team introduced Qwen2- Math. This new language model has more potential to solve arithmetic and mathematical problems as compared to other competitors in this field.
One of the specialties of Qwen2- math is the large amount of resources, books, web texts, code, exam questions. Evaluation on both English and Chinese Mathematical benchmarks are possible. It consists ofGSM8K, Math, MMLU-STEM, CMATH.
“Qwen2- Math instruct achieves the best performance among models of the same size, with RM @ 8outperforming Maj@ 8, particularly in the B and 7B models” the Qwen team said.
Qwen2 showed a high rate of success in American Invitational Mathematics Examination (AIME) 2024 and the American Mathematics Contest (AMC) 2023. Currently the Qwen team is planning to widen itsproficiency beyond English with more multilingual models which will benefit global audiences.
