Core Viewpoint - The Ministry of Ecology and Environment emphasizes the need for strict actions against illegal dumping and disposal of solid waste, focusing on comprehensive inspections and rectifications of related issues [1] Group 1: Illegal Dumping and Waste Management - A special rectification action will be launched to address illegal dumping and disposal of solid waste, with a focus on identifying and correcting these issues [1] - The ministry aims to crack down on illegal dumping behaviors and the profit chains behind them [1] - There will be a thorough investigation into environmental pollution risks at landfill sites, particularly concerning leachate accumulation and groundwater contamination [1] Group 2: Long-term Management Strategies - The ministry plans to analyze the root causes of these issues and develop fundamental solutions to prevent future occurrences [1] - A long-term mechanism for comprehensive solid waste management will be established to ensure ongoing prevention of pollution problems [1]

中心创新推出多项全国首创功能：全国首个"人口信息查询"通道，破解当事人"查人难"痛点；全省首 推"AOL授权见证通"，将港澳当事人授权见证时间从30天压缩至5分钟；首创"涉港澳案件线上授权见 证"，支持跨境授权委托全流程在线办理。此外，中心还与"粤省事""穗好办"等政务平台对接，提供简 体、繁体中文及英语、葡萄牙语四语诉讼指引，助力大湾区法治深度融合。 南都讯记者刘嫚发自北京 6月11日，中国社会科学院法学研究所发布2025年《法治蓝皮书·中国法院信息 化发展报告》（以下简称《法院信息化蓝皮书》）。《法院信息化蓝皮书》显示，针对送达难题，广州 中院推出的AOL电子诉讼（服务）平台创新"送必达"机制，通过电子推送、短信提醒等功能，缩短送达 周期超60%。 《法院信息化蓝皮书》显示，广东省广州市中级人民法院正式推出"AOL电子诉讼（服务）中心"(以下 简称"AOL中心"），通过深度融合前沿技术与司法需求，构建起全流程在线诉讼服务体系。该平台 以"全业务线上办理、全流程依法公开、全方位智能服务"为核心目标，致力于破解传统司法服务中的时 空限制与效率瓶颈，为粤港澳大湾区法治建设注入新动能。 《法院信息化蓝皮书》介绍， ...

"可享受国家补贴""国补再减15%"——这样的宣传标语在2025年的消费市场已成为寻常风景。据商务部 数据，截至5月31日，2025年消费品以旧换新五大品类合计带动销售额1.1万亿元，发放直达消费者的补 贴约1.75亿份。这组数据背后，是一场关于财政资金使用效率的革命性探索正在消费领域展开，其核心 在于打破传统补贴模式的层级壁垒，让政策红利直达民生末梢。 纵观政策设计，中央与地方的财政协同机制展现出精准的制度创新。根据政府工作报告部署，中央通过 超长期特别国债安排3000亿元专项资金，较上年增加1500亿元，截至目前已有1620亿元分两批下达地 方。这种资金安排并非简单"撒胡椒面"，而是建立在对区域差异的精准把握上：东部地区中央承担 85%、中部90%、西部95%，既确保政策全国覆盖，又体现对欠发达地区的倾斜支持。地方财政则严格 按比例配套，形成稳定的资金池。这种"中央定标准、地方保落实"的协作模式，从根本上解决了以往惠 民政策"中央热、地方冷"的落实难题。 政策效能的最大化，关键在于资金流转机制的再造。山东省财政厅披露的预拨机制颇具代表性，该省允 许企业提前申领80%补贴资金，彻底改变了过去"企业先垫付、财 ...

编号：2025-03 | | ☑特定对象调研 □分析师会议 | | --- | --- | | 投资者关系 | □媒体采访 业绩说明会 | | 活动类别 | □新闻发布会 □路演活动 | | | □电话会议 □其他 | | | 中泰证券、中国人保资管、太平洋资管、中加基金、英大 | | | 资产、太平养老、永赢基金、上银基金、万家基金、伟晟 | | | 投资、同泰基金、米利都私募、太平基金、东方证券、英 | | | 大保险资管、中国中信金融资管、海富通基金、诺德基金、 | | | 广发基金、大成基金、中信建投基金、国联安基金、华商 | | 参与单位 | 基金、华泰资产、中银基金、重阳投资、平安资管、泰信 | | | 基金、路博迈基金、博时基金、中国人寿养老、中金汇理 | | | 资管、天弘基金、中国人寿资管、浦泓私募、诺德基金、 | | | 华商基金、深圳创华投资、深圳德讯证顾、幸福阶乘基金、 | | | 昭华基金、亿峰资产、个人投资者杨彪、国泰海通证券 | | 时间 | 2025.5.30、6.03、6.05 | | 地点 | 深圳、上海 | | 上市公司 | 信披与投关工作小组成员 | | 接待人员 | ...

Core Viewpoint - The company intends to divest its wholly-owned subsidiary, Yihui Technology, due to ongoing losses impacting overall performance and negative factors affecting its reputation, with the stake being acquired by Cangnan State-owned enterprise, Shanhai Data Technology [2][4][5]. Group 1: Financial Performance - The parent company, Sichuang Yihui, has faced continuous losses since 2021, with net profits of -685 million yuan, -878 million yuan, -874 million yuan, and -502 million yuan projected from 2021 to 2024 [3]. - Yihui Technology, the core of the smart medical segment, reported a revenue of 169 million yuan in 2024, down significantly from 417 million yuan in 2022, with a net profit of -320 million yuan [3][4]. - The smart medical business has been severely impacted by market competition and administrative penalties, leading to substantial operational losses [4]. Group 2: Administrative Issues - The company faced administrative penalties due to a fraudulent issuance incident, resulting in a fine of 85.7 million yuan and a 10-year market ban for the former chairman [4][5]. - The negative reputation from these penalties has hindered Yihui Technology's ability to secure contracts with major public hospitals, affecting its business development [4]. Group 3: Strategic Shift - The divestiture of Yihui Technology is expected to improve the company's financial situation, with the estimated valuation of the subsidiary at approximately 300 million yuan [5]. - Following the sale, the company will focus on its commercial intelligence business, which has shown a revenue decline of 15.01% in 2024, generating 501 million yuan [7][8]. - The commercial intelligence segment is a significant player in the EAS market and aims to leverage its position to recover from current challenges [8].

Core Insights - Huawei has launched a new model called Pangu Ultra MoE, which has a parameter scale of 718 billion, marking a significant advancement in the MoE model training field on the Ascend AI computing platform [1][3][6] - The release of Pangu Ultra MoE and the Pangu Pro MoE series demonstrates Huawei's capability in achieving a fully controllable training process for domestic computing power and models, validating the innovation capacity of China's AI infrastructure [3][6] Model Architecture and Training Innovations - The Pangu team has introduced innovative designs in model architecture and training methods to address the challenges of training ultra-large-scale and highly sparse MoE models, achieving stable training on the Ascend platform [1][4] - Key innovations include the Depth-Scaled Sandwich-Norm (DSSN) architecture and TinyInit initialization method, which have enabled long-term stable training with over 18TB of data [4] - The introduction of the EP loss load optimization method ensures better load balancing among experts and enhances their specialization capabilities [4] Performance and Efficiency Improvements - The training methods disclosed by Huawei have enabled efficient integration of large sparse MoE reinforcement learning (RL) post-training frameworks on the Ascend CloudMatrix 384 supernodes [5] - Recent upgrades have improved the pre-training system's performance, increasing the multi-factor utilization (MFU) from 30% to 41% [5] - The Pangu Pro MoE model, with 72 billion parameters and 16 billion active parameters, has demonstrated performance comparable to larger models, ranking first among domestic models under 100 billion parameters in the SuperCLUE leaderboard [5] Industry Implications - The successful training and optimization of ultra-large-scale sparse models on domestic AI platforms signify a closed-loop of "full-stack domestication" and "fully controllable processes" from hardware to software, and from research to engineering [6] - This advancement provides a strong foundation for the development of China's AI industry, reinforcing confidence in domestic AI capabilities [3][6]

5月30日，券商中国记者从华为获悉，华为在MoE模型训练领域再进一步，重磅推出参数规模高达7180亿的全 新模型——盘古Ultra MoE，这是一个全流程在昇腾AI计算平台上训练的准万亿MoE模型。同时，华为发布盘 古Ultra MoE模型架构和训练方法的技术报告，披露众多技术细节，充分体现了昇腾在超大规模MoE训练性能 上的跨越。 中国AI产业注入强心剂。 业内人士分析，华为盘古Ultra MoE和盘古Pro MoE系列模型的发布，证明华为不仅完成了国产算力+国产模 型的全流程自主可控的训练实践，同时在集群训练系统的性能上也实现了业界领先。这意味着国产AI基础设 施的自主创新能力得到了进一步验证，为中国人工智能产业的发展提供了一颗"定心丸"。 国产算力与国产模型重大突破 据悉，训练超大规模和极高稀疏性的MoE模型极具挑战，训练过程中的稳定性往往难以保障。针对这一难题， 华为盘古团队在模型架构和训练方法上进行了创新性设计，成功地在昇腾平台上实现了准万亿MoE模型的全流 程训练。 在模型架构上，盘古团队提出Depth-Scaled Sandwich-Norm（DSSN）稳定架构和TinyInit小初始化的方法， ...

现在，请大家一起 数一下"1"、"2" 。 OK，短短2秒钟时间，一个 准万亿MoE大模型 就已经吃透如何解一道 高等数学大题 了！ 而且啊，这个大模型还是 不用GPU 来训练，全流程都是大写的 "国产" 的那种。 这，就是华为通 过 " 昇腾+Pan gu Ultra MoE" 这套 组合拳解锁的效果—— 不仅实现了国产算力与国产模型全流程自主可控的训练闭环，更是在集群训练系统性能方面达到行 业领先水平。 有多领先？来看一组数据： 预训练阶段：昇腾Atlas 800T A2万卡集群MFU提升至41% 后训练阶段：单CloudMatrix 384超节点吞吐达35K Tokens/s 值得一提的是，华为还 首次 把背后的一大秘籍给亮了出来。 具体来说，华为在这次发布的 技术报告 中，披露了在昇腾CloudMatrix 384超节点上，高效打通 大稀疏比MoE强化学习后训练框架的关键技术。 此举可以说是让以强化学习（RL）为核心机制的后训练，进入到了超节点集群时代。 不用GPU的"炼"准万亿大模型方法 在深入华为Pangu Ultra MoE训练系统全流程之前，老规矩，我们还是先来了解一下此前的技术 痛点。 整体 ...

社会经济研究员Chey Tech指出，这套体系实现了"内外双控"：在内部，农户之间将形成互相监督 的生产规范机制；在外部，消费者和贸易商可通过二维码验证产品的原产地真实性，确保产品质量和信 誉。 邓迪那大臣在考察期间深入了解了菩萨省奥松乡榴莲种植户的生产情况及其积极拓展中国市场的努 力。一位拥有30公顷种植规模的农户表示，其年利润已达25万美元。此外，当地另有农户正在积极建设 现代化榴莲包装厂，为即将到来的对华出口做好充分准备。 邓迪那在考察中呼吁广大农户务必坚持良好农业规范（GAP）。同时，他承诺农林渔业部将提供全 方位的支持，包括农场注册、农药安全使用指导以及出口设施认证等，以确保柬埔寨农产品在国际市场 上具备更强的竞争力。 （原标题：柬埔寨推行农产品"原产地识别二维码标签"计划 赋能农业出口） 柬埔寨农林渔业部正大力推行一项创新举措，旨在为本土农产品引入"原产地识别二维码标签"系 统。此举旨在有效遏制假冒伪劣产品，并全面提升消费者对柬埔寨农产品的信心。 农林渔业部大臣邓迪那于上周考察菩萨省榴莲种植园时，正式对外宣布了这项重要计划。据介绍， 该二维码不仅作为数字化标识，更将承载全流程溯源信息，涵盖农产品的 ...

Core Insights - Fumin Bank is focusing on building competitive digital products and distinctive digital enterprise services to enhance its financial ecosystem and drive innovation through technology [1][5] - The bank has achieved significant milestones, including total assets of 62.3 billion yuan and serving over 50 million customers, while also being recognized in various industry rankings [1] Group 1: Digital Financial Services - Fumin Bank has launched the "Arrival Financing" product to alleviate financing difficulties for small and medium-sized suppliers by converting their credit into transaction credit based on delivery scenarios [2][4] - The bank collaborates with a well-known industrial e-commerce platform, which had a total transaction volume exceeding 100 billion yuan in 2024, to provide comprehensive financial services to small suppliers [2][4] Group 2: Efficient Loan Processing - The bank offers a fully online supply chain financing service, allowing suppliers to apply for loans and receive approval and funding within a day, significantly improving efficiency [3] - A case study highlights a supplier receiving a credit limit of 10 million yuan and successfully securing funding the day after delivery, demonstrating the effectiveness of the "Arrival Financing" product [3] Group 3: Ecosystem Development - Fumin Bank aims to enhance its business capabilities by integrating technology, business, and ecosystem to support high-quality development [4] - The bank is committed to providing comprehensive supply chain financial solutions across various industries, including commodities, automotive, steel, and construction, to address the financing challenges faced by small and medium-sized enterprises [4] Group 4: Future Directions - The bank plans to explore new models and products for serving the digitalization of industries, contributing to the transformation and upgrading of traditional sectors [5]