Core Viewpoint - The forum held at Renmin University focused on enhancing ideological and political education in universities, emphasizing collaboration between Beijing and Shanghai institutions to implement the "New Era Moral Education Project" effectively [1][2]. Group 1: Forum Highlights - Nearly a hundred university student work department heads from Beijing and Shanghai discussed practices and experiences related to building practical education classrooms, online education platforms, and collaborative education communities [1]. - Key universities such as Peking University, Tsinghua University, Shanghai Jiao Tong University, and Fudan University participated in the discussions [1]. Group 2: Policy and Implementation - The Beijing Municipal Education Commission emphasized the need for a comprehensive development of ideological courses, creating a "Four Centers" matrix brand for ideological education, and enhancing the capabilities of counselors [1]. - The Shanghai Municipal Education and Health Work Committee highlighted the integration of ideological education across all educational levels, focusing on reforming and innovating mental health education and strengthening the ideological work team [1]. Group 3: National Education Strategy - The Deputy Director of the Ministry of Education's Ideological and Political Work Department stated that the New Era Moral Education Project is a top priority for building a strong educational nation, requiring systematic planning and implementation of key tasks [2]. - Emphasis was placed on creating a comprehensive educational framework that includes innovative theoretical courses, practical education classrooms, and digital education platforms [2].

近日，一位名为"中央音乐学院考级办杨老师"的网友，在网络上宣传2025"艺数未来"数字教育艺术成果展示展演活动，引发关注。1月8日，中央音乐学院紧 急发声，自始未对此人及上述活动进行过任何授权。 多所高校发声 澄清这些"坑" 寒假临近，"教授内推""直通名校""招生咨询"等大量所谓"内部消息"开始在网络上悄然流传。多所高校接连发表官方声明，对各类虚假招生信息、违规活 动、违规使用学校名称及标识等行为予以辟谣。 香港大学 2025年12月30日，香港大学通过"港大招生"发表声明称，发现有不法中介声称可"保录"港大并出示伪造录取通知书，经查均属虚假。港大从未授权任何内地 中介招生，亦不存在"教授内推"等非正规渠道。 清华大学 2025年12月24日，清华大学招生办发表声明称，接到多所中学、家长及学生反映，有部分机构与个人冒用该校名义开展招生宣讲，散布不实招生政策，造成 不良社会影响。清华大学从未授权或委托任何个人及机构举办任何形式的与招生挂钩的营利性培训、宣讲或咨询活动。 北京大学 2025年12月16日，北京大学物理学院发表声明称，有社会机构、个人冒用北京大学物理学院名义开展"北京大学'寒假学堂'营"的活动。北 ...

本文转自【人民日报】； 军事素质稳居前三 学业方面飞行学员班级第二 会跳伞、能主持 最近 清华"飞班"的陈秀虎同学 得到了很多网友的关注 飞行学员陈秀虎 获评清华学生年度人物 清华大学航天航空学院 2023级本科生 陈秀虎 是清华大学与空军航空大学 联合培养的飞行学员 2025年12月31日 清华2025学生年度人物揭晓 陈秀虎榜上有名 此前 他还入围了清华特奖的答辩 清华大学的特等奖学金 是清华授予在校生的最高荣誉 清华特奖被网友称为"神仙打架" "能入围的全都是 学霸中的'战斗机'" ▲参加清华特奖评选时的简介 自幼立志从军报国 他终圆蓝天梦 陈秀虎 从小就立志从军报国 高中时的陈秀虎 高中时 陈秀虎就读于绵阳中学航空实验班 曾荣获"绵阳市三好学生" 初中时 他就参加了招飞体检 2023年7月 他以第一名8.98的招飞心品测试成绩 （满分9分） 良好的身体素质和扎实的文化基础 通过空军招飞 进入清华大学 航天航空学院飞行学员班 成为"双学籍"飞行学员 大学时期的陈秀虎 在清华、北大、北航三校飞班的 两次大比武中稳居前三名 跳伞考核中 遭遇主伞未能正常张开的特情 他临危不惧，成功处置 顺利完成跳伞 学业方面 ...

撰文丨王聪 编辑丨王多鱼 排版丨水成文 尽管药物研发领域取得了诸多进展，但仍有约 90% 的可成药疾病靶点缺乏小分子药物。随着诸如 AlphaFold 等蛋白质结构预测技术的进步， 全基因组药物发现 已成为一个更可实现的目标。然而，目前使 用的 虚拟筛选 （ Virtual Screening ） 工具远不能满足这一需求。现有的方法 （无论是经典的分子对接还 是深度学习方法） ，仍存在着计算成本都太高、无法覆盖全基因组靶点等问题。 因此，研究人员希望能够开发出 一种有效的 全基因组虚拟筛选 方法，以快速识别人类基因组中每个可成 药靶点的小分子配体。 如今，这一局面正被打破。来自 清华大学 的研究团队推出了 AI 驱动的超高通量药物虚拟筛选平台—— DrugCLIP ， 首次实现 全基因组规模的虚拟筛选 ，将传统方法需数年的计算任务压缩至 24 小时内，效率 提升最高达 1000 万倍。 该研究于 2026 年 1 月 8 日，清华大学智能产业研究院 （ AIR ） 兰艳艳 教授 联合清华大学生命科学学 院学院 张伟 副教授、 闫创业 副教授及化学系 刘磊 教授 （ 贾寅君 、 高博文 、 谭佳 鑫 、 郑济青 ...

Core Viewpoint - The article discusses a significant breakthrough in AI-driven drug discovery through the development of DrugCLIP, a platform that can perform high-throughput virtual screening of drugs at a genomic scale, achieving 10 trillion protein-molecule pairing calculations within 24 hours [1][4][36]. Group 1: DrugCLIP Platform - DrugCLIP is an AI-driven ultra-high-throughput virtual screening platform developed by Tsinghua University, which allows for rapid identification of candidate drug molecules from vast chemical libraries [2][3]. - The platform has successfully completed virtual screening covering the human genome scale, identifying potential drug molecules for diseases such as depression, cancer, and Parkinson's disease [6][54]. Group 2: Challenges in Traditional Drug Screening - Traditional drug screening faces three main challenges: slow processing speed, lack of starting points for many disease-related proteins, and a narrow focus on popular targets [8][12][18]. - Only 10% of protein targets have mature drugs available, while 90% remain without identified drugs [11]. Group 3: Methodology of DrugCLIP - DrugCLIP employs a novel approach by using contrastive learning to train AI encoders that create vector representations of protein binding pockets and chemical molecules [20][22]. - The model processes 5 billion candidate molecules, generating vector representations to quickly identify the most promising candidates for new drug development [32][34]. Group 4: Performance and Validation - DrugCLIP has demonstrated superior performance in virtual screening benchmarks, outperforming traditional docking tools and other AI methods in identifying effective molecules [37][39]. - Experimental validation showed that from 78 screened molecules related to depression, 8 were found to activate the target protein, with the best molecule exhibiting a binding affinity of 21 nM [42][43]. Group 5: Future Prospects - The DrugCLIP platform is set to collaborate with industry partners to accelerate the discovery of new drug targets and first-in-class drugs for various diseases [64]. - The database created by DrugCLIP, which is now open to the global research community, represents the largest known protein-ligand screening database, potentially providing "drug seeds" for nearly half of human proteins [55][59].

Core Insights - The MARSHAL framework, developed by Tsinghua University and other institutions, utilizes reinforcement learning for self-play in strategy games, significantly enhancing the reasoning capabilities of large models in multi-agent systems [2][7][31] - The framework addresses two main challenges in multi-agent systems: credit assignment in multi-round interactions and advantage estimation among heterogeneous agents [5][7] Background and Challenges - Existing models like DeepSeek-R1 have shown the value of verifiable reward reinforcement learning (RLVR) in single-agent scenarios, but its application in complex multi-agent interactions is still in exploration [5] - The two core technical challenges identified are: 1. Credit assignment in multi-round interactions, where existing methods struggle to accurately trace back results to specific actions [5] 2. Advantage estimation among heterogeneous agents, which complicates joint training and leads to performance volatility [7] MARSHAL Method Introduction - MARSHAL employs Group-Relative Policy Optimization (GRPO) architecture and introduces two key algorithmic improvements to enhance multi-agent reasoning capabilities [12][14] - The framework was tested using six strategy games, with three for training and three for testing, covering a range of competitive and cooperative scenarios [12] Core Experiments - The MARSHAL-trained expert agents demonstrated a significant performance increase, achieving up to 28.7% higher win rates in testing games [13][19] - The model showed remarkable generalization capabilities, with accuracy improvements of 10.0% in AIME and 7.6% in GPQA across various reasoning tasks [19][20] Reasoning Mode Analysis - Qualitative analysis revealed that the training in games fostered two emergent capabilities: Role-Awareness and Intent Recognition, which are crucial for decision-making in uncertain environments [22] - Quantitative analysis indicated that MARSHAL reduced inter-agent misalignment by 11.5%, enhancing communication efficiency among agents [24] Ablation Studies - Self-play training outperformed fixed opponent training, as models trained against fixed opponents tended to overfit, leading to poor performance in testing scenarios [26] - The necessity of the Turn-level Advantage Estimator and Agent-specific Advantage Normalization was confirmed, highlighting their importance in handling long-sequence decisions and addressing reward distribution differences [28] Conclusion - The MARSHAL framework successfully enhances the reasoning capabilities of large language models in multi-agent systems through self-play in strategy games, indicating potential for broader applications in complex multi-agent environments [31][34]

编辑丨王多鱼 排版丨水成文 热带森林 储存了全球约一半的 地上生物量碳 （AGC） ，但广大区域正遭受干扰影响，包括农业扩张导致 的毁林以及火灾、选择性采伐和边缘效应引发的退化。随着时间的推移，受干扰森林能够恢复，逐步重建 碳储量和生态功能。然而，关于恢复速率如何随干扰规模、类型和地理位置变化的问题仍缺乏量化研究。 2026 年 1 月 7 日，清华大学 李伟 副教授 、 法国气候与环境科学实验室 Philippe Ciais 教授教授 作为通 讯作者 （ 徐伊迪 为论文第一作者 ） ，在国际顶尖学术期刊 Nature 上发表了题为： Small persistent humid forest clearings drive tropical forest biomass losses 的研究论文。 该研究基于高分辨率的森林扰动和生物量数据，构建了格点尺度的森林扰动植被恢复数据库，并在此基础 上开发了一个集成高分辨率遥感数据和森林扰动恢复数据库的森林碳簿记模型。利用该模型估算了 1990- 2020 年间森林扰动导致的植被碳储量时空变化动态，明确了不同干扰类型和扰动斑块大小对森林植被碳储 量及碳密度变化的具体 ...

Core Insights - The collaboration between Gaode Map and Tsinghua University has led to the launch of China's first AI smart campus project, "Friendly Community," which aims to enhance campus traffic management and improve the travel experience for students and faculty [1][2]. Group 1: Project Features - The project introduces a campus bus navigation feature that allows users to input start and end points within the campus to receive real-time bus routes, stop locations, and estimated arrival times, significantly reducing wait times [1]. - AI technology has been utilized to address issues such as road congestion during peak hours and difficulty in locating points of interest, with over 120 campus roads and traffic rules accurately mapped [2]. - The platform has optimized nearly 1,600 points of interest (POIs) within the campus, improving the efficiency of locating classrooms, offices, and dining facilities [2]. Group 2: User Engagement and Statistics - Within three months of launch, the project recorded over 100,000 searches for new POIs, nearly 1,000 navigation uses, and a daily search volume of 30,000 for campus buses [2]. - The project is positioned as a model for digital governance in higher education, with expectations for future enhancements such as Gaode's VIP ride-hailing service for Tsinghua students, offering benefits like priority booking and dynamic discounts [2][3]. Group 3: Technological Advancements - The project has completed 3D modeling technology validation, paving the way for future applications in 3D navigation and emergency simulations, thus expanding the scope of digital management on campus [3]. - The partnership between Gaode and Tsinghua not only optimizes campus travel experiences but also establishes a sustainable model for collaboration between enterprises and educational institutions, potentially accelerating the digital transformation of more universities [3].

Group 1 - The core viewpoint of the news is the launch of "Fenghe," China's first generative AI meteorological service language model, which provides personalized and intelligent weather-related services [1] - Fenghe has a high-quality meteorological service corpus of 50 million tokens and a question-answer instruction set of 490,000, offering various data interfaces and personalized tools for sectors like energy, transportation, and tourism [1] - The AI model is already being applied in various meteorological departments, including its integration with the "Sui Xiao Tian" mini-program during the 15th National Games in Guangzhou [1] Group 2 - The development of Fenghe involved breakthroughs in four key technologies: corpus construction, knowledge enhancement and scene fine-tuning, deep reasoning, and multi-agent collaboration [2] - The corpus construction technology converts vast meteorological data into a specific format for the model to learn, while the fine-tuning technology enhances the model's ability to understand user needs and execute complex tasks [2] - The team faced challenges during the development process but remained committed to innovation and recognized the need for further technical support to embrace the AI era in meteorological services [2] Group 3 - Looking ahead to the 14th Five-Year Plan, the training, reasoning, and iteration of AI models will require stronger computing capabilities and large-scale high-quality corpora [3] - There is a need to continuously enhance learning, highlight advantages, and refine the application of intelligent agents in various scenarios to promote the deep integration of "AI+" in meteorological services [3]

1月8日，中央音乐学院通过官方微信公众号发布声明：日前，我校发现有微信用户"中央音乐学院考级 办杨老师"(微信号:BG151108)在未经我校授权情况下，以我校名义宣传2025"艺数未来"数字教育艺术成 果展示展演活动，我校在此郑重声明如下：上述活动与我校无任何关系，我校不对上述活动承担任何责 任，亦自始未对微信用户"中央音乐学院考级办杨老师"（微信号:BG151108）及上述活动进行过任何授 权。 近日，北京大学、清华大学、西南交通大学等多所高校接连发布声明，对各类虚假招生信息、违规活 动、违规使用学校名称及标识等行为进行辟谣。 最新一例来自中央音乐学院。 2025年12月9日，西南交通大学发布声明称，该校正在开展学校名称授权使用的专项清理工作，发现部 分社会企业擅自使用"西南交通大学""西南交大"或使用带有特定指向该校的校名组合字样作为企业名 称，误导公众认为其与西南交通大学存在关联关系。学校称，上述行为已严重侵害该校合法权益，应当 承担由此产生的全部法律责任。 2025年12月19日，复旦大学终身教育处发布声明称，近期，有社会机构、个人冒用复旦大学名义散布研 学营、冬令营等虚假信息开展违规招生。根据教育 ...