Search
Lixiang Tongxue Integrates Advanced DeepSeek-R1-0528 Model
2 day ago /
Read about 0 minute
Author：小编
Lixiang Tongxue's mobile and web platforms have seamlessly integrated the latest iteration of the DeepSeek model, R1-0528. This version stands out in domains including mathematics, programming, and logical reasoning, aligning with international top-tier benchmarks. Its complex reasoning capabilities have undergone a remarkable improvement, achieving an AIME test accuracy rate of 87.5% while reducing the hallucination rate by 45-50%. Furthermore, the new model excels in creative writing, capable of producing long-form pieces with coherent structures and styles that closely mirror human authorship.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
