In July, Baidu's Basic Model Unit (BMU) ushered in its new leader, Sun Tianxiang, who promptly restructured the organizational framework and refined the technical blueprint for ERNIE. On the one hand, BMU recruited Wu Qin, a prominent large-scale model researcher from a state-of-the-art laboratory in North America. With extensive experience in developing multiple generations of large-scale models, Wu is poised to bolster the pre-training capabilities of ERNIE's large-scale models. On the other hand, Wei Haoran, a former researcher at DeepSeek who joined Baidu earlier this year, transitioned with his team to BMU. He was appointed as the head of ERNIE's multimodal algorithm team, bringing with him remarkable accomplishments in large-scale models and OCR technologies. Furthermore, upon assuming the role of BMU head in July, Sun Tianxiang also became a member of Baidu's Model Committee. As a technical expert born in the 1990s, he was selected to join Baidu's Technical Strategy Committee following its reorganization at the end of July.