Group 1 - The core viewpoint of the article is that Jiangsu Province aims to leverage digital trade to promote high-quality development of service trade, with a target of reaching a service trade scale of 600 billion yuan and digital delivery service trade of 300 billion yuan by 2030, accounting for approximately 50% of the service trade [1] - Jiangsu will focus on institutional openness in digital trade, creating a digital trade ecosystem, and aligning with high-standard economic and trade rules, including pilot cooperation in digital trade with Singapore [1] - The province plans to establish national service trade innovation development demonstration zones and national digital trade demonstration zones, enhancing infrastructure and public services in key areas like Nanjing Software Valley to facilitate domestic and international industrial chain collaboration [1] Group 2 - A significant highlight of the policy is industry empowerment, with Jiangsu focusing on developing digital product trade in the cultural industry, strengthening cultural trade bases in cities like Nanjing, Wuxi, and Suzhou, and promoting exports in sectors such as animation and film [2] - The province aims to expand digital technology trade in advantageous fields, advance high-end software development, and implement an "Artificial Intelligence+" action plan to upgrade service outsourcing and promote enterprise transformation [2] - Jiangsu will enhance international transportation service capabilities, optimize international route networks, and accelerate the development of smart ports and waterways, while also improving the international competitiveness of tourism services and supporting international education services [2]

Summary of Conference Call Notes Company and Industry Involved - **Company**: Meta - **Industry**: AI and Technology, specifically focusing on large models and machine learning Core Points and Arguments 1. **Talent Acquisition**: Meta is aggressively recruiting talent from companies like OpenAI, Google, and Anthropic, focusing on areas such as multimodal processing and post-training to enhance the competitiveness of its LLAMA model [1][9][10] 2. **Impact of Talent Loss on OpenAI**: Key members of OpenAI's O1 model team, including Ren Hongyu, Zhao Shengjia, and Yu Jiahui, have left, which has prompted OpenAI to accelerate its development pace [1][12] 3. **AI Talent Salary Surge**: Salaries for top AI talent have skyrocketed, with annual compensation reaching up to $100 million, indicating fierce competition among tech companies for AI professionals [1][11] 4. **Shift in AI Development Strategy**: By the second half of 2025, tech companies will return to the pre-training phase, with Meta focusing on data, Google optimizing architecture, and OpenAI continuing its large cluster strategy [1][29][30] 5. **Increased Demand for AI Computing Power**: The new round of AI innovation is expected to significantly increase the demand for computing power, training, and cluster needs [3][38] 6. **Meta's Role as a Catalyst**: Meta's actions are accelerating changes in the U.S. AI industry, making it a focal point for investment in the coming months [5][38] 7. **Challenges Faced by Meta**: Meta's LLAMA4 model has underperformed, leading to a strategy shift that includes talent acquisition to improve its competitive position [6][19] 8. **Strategic Focus on Data Quality**: Meta's strategy involves acquiring Skill AI to enhance data filtering capabilities, addressing the challenge of extracting valuable insights from vast amounts of data [14][31] 9. **Future of AI Models**: The next generation of models will require significant human resources and computing power, with a focus on capital expenditures to ensure adequate resources for training [39][40] Other Important but Possibly Overlooked Content 1. **Meta's Historical Context**: Meta's journey in AI began in 2013, coinciding with significant industry milestones, and has evolved through various acquisitions and strategic shifts [15][17] 2. **Comparison with Competitors**: While Meta is making strides, it currently lacks globally leading experts in large models, which may hinder its competitive edge [19][20] 3. **Long-term Industry Evolution**: The AI industry has evolved from CNN to RNN and now to Transformer architectures, with ongoing debates about the path to AGI [21] 4. **Investment in Computing Resources**: Companies like OpenAI and XAI are also expanding their computing resources, with OpenAI planning a $30 billion order with Oracle to support its million-card cluster by 2027 [34][33] 5. **Meta's Potential for Growth**: Meta's recent actions may elevate its position in the AI landscape, potentially allowing it to compete more closely with OpenAI and XAI in the next model iteration [25][36]

Core Viewpoint - The article discusses the legal implications of using copyrighted works as training data for AI models, advocating for a "wide entry, strict exit" approach to balance AI development and copyright protection [1][2][3]. Group 1: Legal Framework for AI Training Data - The author suggests establishing a reasonable use system for AI training data at the "input end" while implementing stricter regulations at the "output end" to protect the interests of copyright holders [1][2]. - The current risks associated with AI model applications are unclear, and imposing strict regulations at the input stage could hinder innovation due to high authorization costs and legal risks for AI developers [2][3]. - The author argues that traditional copyright licensing models may suppress innovation due to high costs and complex negotiations, leading to potential legal gray areas for AI companies [2][3]. Group 2: Legislative Recommendations - The author recommends legislative measures to classify AI training data as a specific case of reasonable use under copyright law, emphasizing its public interest and value in the AI industry [3]. - The use of training data by AI models is compared to "molecular gastronomy," where the data is not merely copied but transformed to extract underlying patterns [3]. - The proposal includes providing copyright holders with remedies for legal data acquisition and infringement risks, ensuring a dynamic balance between reasonable use and copyright protection [3]. Group 3: Judicial Precedents - Recent U.S. court rulings on AI training data have significant implications for China, highlighting the need for careful examination of whether the use of copyrighted works negatively impacts their market value [4]. - The rulings indicate that while some uses may be deemed reasonable, the legality of using copyrighted works for training AI models remains a complex issue that requires case-by-case analysis [4].

Core Viewpoint - The next significant leap in semiconductor packaging will require a series of new technologies, processes, and materials that will collectively achieve an order-of-magnitude performance improvement, which is crucial for the AI era [1]. Group 1: Advances in Cooling Technologies - Liquid cooling technology at the chip level is emerging as forced air cooling reaches its limits, with up to 40% of power used for current delivery and heat dissipation [4]. - TSMC's silicon integrated micro-cooler (IMEC-Si) is being tested for reliability, designed to handle over 3,000 watts of uniform power dissipation under specific conditions [6]. - The demand for direct liquid cooling is increasing, with innovative concepts like using chips as coolants being proposed [7]. Group 2: Hybrid Bonding and Interconnects - Hybrid bonding with fine-pitch multilayer redistribution layers (RDL) is gaining attention as a cost-effective solution for high-speed interconnects [14]. - Intel's hybrid bonding can achieve spacing as small as 1µm, which is critical for advanced applications [5][17]. - The transition from traditional dielectric materials to polymer/copper hybrid bonding is being explored to enhance performance [16]. Group 3: Backside Power Delivery - Backside power delivery significantly reduces voltage drop related to transistor power supply, but it also exacerbates heat issues [19]. - IBM has developed an anisotropic model for precise heat transfer calculations in backend stacks, emphasizing the importance of thermal considerations in design [21]. - The implementation of backside power delivery is expected to lead to a 10% to 30% reduction in thermal losses [23]. Group 4: Co-Packaged Optical Devices - The demand for faster data networks is driving the integration of optical engines with GPUs and HBM in a single package, significantly increasing data transmission speeds [26]. - Co-packaged optical devices (CPO) are expected to achieve a 32-fold increase in bandwidth by bringing optical engines closer to processors [26]. - However, challenges remain regarding thermal management and warpage sensitivity in CPO implementations [28].

据央视新闻， 近日，海军辽宁舰、山东舰航母编队圆满完成远海实战化训练任务返港。 这次 远海实战化训练，两支航母编队围绕侦察预警、防抗打击、对海突击等多项课题开展研练，此 次训练任务是根据年度计划组织的例行性训练，有效检验了部队联合训练成效，提高了维护国 家主权、安全、发展利益的能力。 海军山东舰编队 高翔： 任务期间，围绕航母编队使命任务，编队开展多项核心作战能力强化 训练，实际检验各项方案预案，有效提升编队作战能力。 任务期间，外军舰机多次抵近侦察、跟踪监视，辽宁舰、山东舰航母编队全程保持高度戒备、 随时反应的作战状态，多次组织舰载机战斗起飞，专业稳妥处置应对，确保了各项训练任务的 圆满完成。 海军山东舰 曾文辉： 此次远海实战化训练，海空域气候条件复杂，训练强度大，对保障舰载 机起降提出了更高要求。全舰各部位密切协同配合，细化方案预案，严把飞行安全观，确保圆 满完成舰载机训飞保障任务。 任务过程中，辽宁舰、山东舰航母编队积极探索编队作战要素和力量实战运用，形成多项课题 研究成果，有力提升航母编队体系作战能力。这是继去年首次组织双航母编队联合演练后，两 支航母编队又一次位远海大洋开展体系对抗演练。 在这次远 ...

Core Viewpoint - The successful completion of the far-sea combat training by China's aircraft carrier groups, including Liaoning and Shandong, enhances the military's capability to safeguard national sovereignty and interests [1] Group 1: Training Overview - The training involved two aircraft carrier groups focusing on reconnaissance, anti-strike, and maritime assault tasks [1] - This training was part of a routine exercise organized according to the annual plan, effectively testing the joint training outcomes of the military [1] Group 2: Operational Readiness - During the training period, foreign military vessels and aircraft conducted multiple reconnaissance and monitoring operations near the training area [1] - The Liaoning and Shandong carrier groups maintained a high state of alert and readiness, conducting multiple combat takeoffs of carrier-based aircraft to respond to potential threats [1]

Core Viewpoint - The ZEEKR autonomous driving team won the Argoverse2 2025 scene mining challenge at the CVPR 2025 conference, showcasing their AI technology's ability to address recognized challenges in the global autonomous driving sector [1] Group 1: Competition Results - The ZEEKR autonomous driving team achieved first place in the Argoverse2 2025 scene mining challenge [1] - The competition was held at the prestigious CVPR 2025 conference, highlighting the significance of the achievement [1] Group 2: Technological Advancements - The team utilized AI technology to effectively solve key technical challenges in the autonomous driving field [1] - Their AI model demonstrated superior learning outcomes from large and redundant datasets, enhancing the system's ability to identify and process critical driving scenarios in real-world applications [1]

Core Viewpoint - The article discusses the advancements in end-to-end autonomous driving technology, highlighting the performance of the top competitor, Poutine, in a recent visual-based driving competition, emphasizing its robust training methodology and superior results [1][13]. Group 1: Technical Overview - The leading solution, Poutine, utilizes a 3B parameter Vision-Language Model (VLM) to address long-tail scenarios in visual end-to-end autonomous driving [1]. - The training process consists of two phases: - Phase one involves self-supervised pre-training using a combination of vision, language, and trajectory data, with a total of 83 hours of CoVLA data and 11 hours of Waymo long-tail dataset [2]. - Phase two focuses on fine-tuning through reinforcement learning (RL) using 500 segments of manually annotated data from the Waymo validation set to enhance robustness [2][8]. - The Poutine model achieved a Rater-Feedback Score (RFS) of 7.99 on the Waymo test set, leading the competition [2][13]. Group 2: Data and Methodology - The datasets used include CoVLA, which contains 10,000 front-view images and 30 seconds of driving video, and WOD-E2E, which provides 4,021 long-tail driving scenarios with trajectory information [11]. - The evaluation metric, RFS, is calculated based on the proximity of predicted trajectories to expert-rated trajectories, with a scoring range of 0 to 10 [11]. - The training details include a batch size of 64 and a learning rate of 1e-5 for the CoVLA dataset, while the WOD-E2E dataset used a batch size of 16 with similar training parameters [11]. Group 3: Results and Analysis - Poutine's performance significantly outperformed other models, with a notable score of 7.99, while the second-best model scored 7.91, indicating a substantial lead [13]. - The article notes that while the addition of RL did not drastically improve scores, it effectively addressed challenging scenarios [13]. - The results suggest that the combination of VLM and RL training enhances the model's ability to handle complex driving environments [18]. Group 4: Future Considerations - The article raises questions about the mainstream applicability of VLM and LLM in trajectory prediction, particularly regarding their understanding of the physical world and 3D trajectory information [19]. - It suggests that for conventional evaluation datasets, the advantages of such models may not be as pronounced, indicating a need for further exploration [19]. - The potential integration of action models with VLM for trajectory prediction is proposed as a more comprehensive approach [19].

点击下方 卡片 ，关注" 具身智能 之心 "公众号 作者丨 Tian-Yu Xiang等 编辑丨具身智能之心 本文只做学术分享，如有侵权，联系删文 >> 点击进入→ 具身智能之心 技术交流群 更多干货，欢迎加入国内首个具身智能全栈学习社区 ： 具身智能之心知识星球 (戳我) ， 这里包含所有你想要 的。 想象学习走路的情景：尽管祖先的经验让一些与生俱来的能力（例如：平衡感、反应）被编码到我们的 DNA中，但要真正学会走路，仍需要在真实环境中不断练习、摔倒、再爬起。经过一段时间的训练，我们 的大脑和身体会逐渐协调一致，形成与环境交互的策略。这种 由通用能力到特定技能 的转变过程在人类中 十分常见，而如今， 智能机器人 也面临着类似的挑战：即便拥有强大的预训练模型作为"大脑"，在执行具 体复杂任务前，仍需要经过类似于人类学习的"后训练"阶段，才能在新环境、新任务下达到理想表现。 1. 概述 这项工作从 人类运动技能学习 的角度系统性地对总结 VLA模型（视觉-语言-动作模型） 的 后训练（post- training）策略 。其主要贡献如下： (1) 从人类运动学习视角讨论了VLA模型后训练方法 ：将人类运动技能 ...

Core Viewpoint - The article discusses the modern challenges of emotional stability in an age of information overload and social media, emphasizing the importance of inner freedom and emotional autonomy as a means to maintain mental health and well-being [1][6]. Group 1: Historical Context - Ancient Chinese scholars, such as Su Shi, have deeply understood and practiced the concept of emotional stability, illustrated by the story of Su Shi and the Zen master Fo Yin, which highlights the difficulty of achieving true emotional detachment despite theoretical understanding [2][3]. Group 2: Psychological Insights - Modern psychology supports the ancient wisdom, with theories like emotional autonomy indicating that mature individuals can distinguish their emotions from others and do not base their self-worth on external evaluations [4]. - Cognitive Behavioral Therapy (CBT) aims to help individuals identify and change negative thought patterns, thereby regaining control over their emotions [5]. - Mindfulness training encourages individuals to observe their emotions without becoming entangled in them, reflecting the ancient ideal of remaining unaffected by external disturbances [5]. Group 3: Contemporary Challenges - The current societal landscape presents unprecedented challenges to emotional autonomy, with social media fostering a performative identity, consumerism linking happiness to material possessions, and information overload fragmenting attention [6]. - The difficulty of maintaining inner peace amidst external chaos is underscored by the quote from Pascal, emphasizing the struggle to be alone and quiet in today's world [6]. Group 4: Strategies for Emotional Stability - Developing awareness is crucial; individuals should observe their emotional responses and identify automatic thought patterns [7]. - Clarifying personal values helps individuals resist societal pressures and maintain their emotional integrity [8]. - Establishing boundaries is essential for protecting one's mental space, allowing individuals to say "no" when necessary [8]. - Pursuing a sense of meaning by focusing on personal growth rather than external validation enhances resilience against emotional fluctuations [8]. - The goal is not to achieve a perfect emotional state but to cultivate the ability to coexist with emotions, acknowledging their presence without being controlled by them [8].