Core Viewpoint - The article discusses the launch of the Xinhua Large Model, a generative AI service developed by Jiangsu Xinhua Media Group, which aims to support theoretical learning for government and media sectors, marking a significant step in the integration of AI in these fields [1][3]. Group 1: Overview of the Xinhua Large Model - The Xinhua Large Model is a vertical AI product designed for theoretical learning, providing specialized and intelligent support for personnel in institutional settings [1]. - As of December 2, 2025, Jiangsu has registered nine generative AI services, including the Xinhua Large Model, following the implementation of the interim management measures for generative AI services introduced in July 2023 [1][2]. Group 2: Company Background and Capabilities - Jiangsu Xinhua Media Group, a modern multimedia group led by the Xinhua Daily, has an asset scale exceeding 10 billion yuan as of the end of 2024, showcasing strong financial and technological capabilities [2]. - The group has established high-quality data sets, including historical newspaper data and smart media resource data, which serve as a foundation for the development of the Xinhua Large Model [2]. Group 3: Technical Features and Services - The Xinhua Large Model is developed based on a fine-tuned domestic general model, utilizing Transformer architecture and PyTorch framework, and offers functionalities such as news creation assistance and theoretical learning Q&A [3]. - It provides five core services: intelligent Q&A for theoretical learning, literature interpretation and analysis, knowledge retrieval and recommendation, support for learning and creation, and personalized learning path development [3]. Group 4: Future Development Plans - The company plans to enhance the Xinhua Large Model by increasing R&D investment to improve service accuracy and exploring integration with new technologies like knowledge graphs [4]. - There is a focus on building a systematic and practical "intelligent writing matrix" to enhance the model's creative capabilities, ensuring it produces content with depth, warmth, and professionalism [4]. Group 5: Industry Impact - The integration of AI in Xinhua Media is seen as a strategic move that creates greater value across various applications, contributing to the overall prosperity of the AI ecosystem [5].

Core Insights - The article discusses the identification of a Chinese pilot, Huang Guanghan, who was the only survivor rescued by John Rabe during the Nanjing Massacre, highlighting the significance of this identification for historical research on the Nanjing Massacre and Chinese aviation history [1][2][3] Group 1: Historical Context - Huang Guanghan served as the squadron leader of the 22nd Squadron of the 4th Group of the Chinese Air Force during the Second Sino-Japanese War, where he shot down multiple Japanese aircraft [5][6] - The identification of Huang Guanghan as "Luo Fuxiang" in Rabe's diary is crucial for understanding the historical narrative of the Nanjing Massacre and the role of Chinese pilots [1][2] Group 2: Research and Verification - Researchers utilized various historical documents, including Rabe's diary and Chinese aviation archives, to confirm Huang Guanghan's identity, establishing a connection between his experiences and the accounts documented in Rabe's writings [2][3] - The collaboration among historians and the analysis of primary sources have led to a consensus on Huang Guanghan's identity, which is seen as a significant contribution to the study of the Nanjing Massacre [2][3] Group 3: Cultural Significance - The article emphasizes the importance of preserving and understanding the narratives of individuals like Huang Guanghan, as they contribute to the broader historical context of wartime experiences and cultural memory in China [1][2] - The recognition of Huang Guanghan's story serves as a reminder of the human experiences behind historical events, fostering a deeper understanding of the impact of war on individuals and communities [1][2]

本报讯(记者聂伟)记者从江苏省委网信办获悉，截至2025年12月2日，我省又新增了新华大模型等9款已 完成备案的生成式人工智能服务。 生成式人工智能服务是指利用生成式AI技术向公众提供生成文本、图片、音频、视频等内容的服务。 我国2023年7月正式出台《生成式人工智能服务管理暂行办法》，明确了对生成式AI大模型研发及服务 的合规要求。此次完成备案的新华大模型生成式人工智能服务备案号为：Jiangsu-XinHua- 202510310055，由江苏新华报业传媒集团有限公司自主研发，是一款面向理论学习领域的垂直型人工智 能产品，其用户群体聚焦党政机关、国有企业及传媒单位，适用场景覆盖思想学习、主题教育等各类理 论学习活动。新华大模型基于国产通用模型微调开发，可提供五大核心服务：一是理论学习领域智能问 答服务，为用户实时解答学习过程中的疑问；二是文献解读与分析服务，助力用户快速把握文献核心要 义；三是知识检索与推荐服务，精准匹配用户所需学习资源；四是辅助学习与创作服务，为学习材料撰 写等提供支持；五是个性化学习方案定制服务，根据用户需求定制专属学习路径。 新华大模型通过"苏大强.爱学习"智能体等技术产品对外提供服 ...