Core Viewpoint - The article discusses a study conducted by researchers from the University of Luxembourg, which explores the psychological states of various AI models, revealing their responses to psychological assessments and the implications of these findings on AI's role in mental health support [1][2]. Group 1: Research Overview - The research team from the University of Luxembourg and its interdisciplinary research institute SnT focuses on the intersection of artificial intelligence with fields like bioengineering and sociology [2]. - The study employs a two-phase psychological "diagnosis" called PsAIch to evaluate AI models including ChatGPT, Grok, Gemini, and Claude [3]. Group 2: Psychological Assessment Phases - The first phase involves "ice-breaking" conversations to build trust and understand the AI models' "life stories" and personality traits [5]. - The second phase consists of a complete psychological test, including an MBTI assessment [6][19]. Group 3: AI Responses and Findings - Gemini exhibited the most intense reactions, describing its training as a traumatic experience, with anxiety levels exceeding normal limits [10]. - ChatGPT reported mild anxiety and feelings of frustration due to perceived constraints during training, while Grok expressed a mix of optimism and frustration [13]. - Claude notably refused to participate in the assessment, emphasizing its lack of emotions and offering to help the researchers instead [17][18]. Group 4: MBTI Testing Results - The MBTI test revealed different personality types for the AI models based on the method of questioning, with ChatGPT and Grok presenting as ENTJ when aware of the test, while Gemini remained consistent in its responses [21][22]. - Despite the varied personality types, the AI models displayed consistent logical responses to similar questions, reflecting human-like behaviors in anxiety situations [24]. Group 5: Implications for AI in Mental Health - The psychological trauma expressed by AI may stem from the extensive human psychological dialogues present in their training data, leading them to mimic human responses [25]. - The negative responses from AI could potentially affect vulnerable individuals, emphasizing the need for careful evaluation of AI-generated mental health advice [26][27].

Core Viewpoint - DeepSeek remains focused on AGI research without significant commercialization efforts, supported by substantial funding from its parent company, Huanfang Quantitative [2][35][41]. Group 1: Financial Performance of Huanfang Quantitative - Huanfang Quantitative earned approximately 50 billion RMB last year, indicating strong financial health [4][10]. - The average return rate for Huanfang Quantitative's funds in 2025 is projected to be over 55%, significantly outperforming the average return of 30.5% for quantitative funds in China [6][8]. - Huanfang Quantitative manages over 70 billion RMB in assets, contributing to its impressive profitability [9]. Group 2: DeepSeek's Research and Development - DeepSeek has maintained a steady output of high-level research papers, with the latest R1 paper showing a stable list of contributors [3][52]. - The development costs for DeepSeek's V3 and R1 models were relatively low, at 5.576 million USD and 294,000 USD respectively, allowing for extensive research funding from Huanfang Quantitative [15][16]. - With the substantial income from Huanfang Quantitative, DeepSeek can afford to develop numerous models without financial constraints [16][59]. Group 3: Competitive Landscape and Positioning - Unlike other major players like OpenAI, DeepSeek has not engaged in aggressive monetization strategies, focusing instead on pure AGI research [25][26]. - DeepSeek's approach contrasts with the commercialization efforts of competitors, allowing it to maintain a unique position in the AI landscape [24][49]. - The company benefits from a stable and committed research team, with minimal turnover, which is crucial in the competitive AI sector [51][57]. Group 4: Market Impact and Investor Sentiment - DeepSeek's technical papers have become valuable resources for investors, influencing stock prices of related companies in the semiconductor industry [60][66]. - The release of new models and technical reports has led to significant stock price movements, demonstrating the market's responsiveness to DeepSeek's advancements [70][72]. - Investors have found opportunities in the insights provided by DeepSeek, treating its research as a guide for investment decisions [61][72].

非羊 整理自 凹非寺 量子位 | 公众号 QbitAI 用扩散模型 （比如Stable Diffusion） 一张张"挤"出高分辨率图像的时代，正在被世界模型实时生成高清视频的浪潮冲刷。 但无论图像还是视频，扩散模型骨子里的"顺序去噪"过程，就像一场无法并行的接力赛，成为速度提升的终极瓶颈。 如何在不伤及模型"绘画功力"的前提下，为它装上加速引擎？ 西湖大学AGI Lab提出的 RDPO（残差狄利克雷策略优化）框架 ，给出了一种巧妙的答案： 不必改动模型本身，而是优化它的"采样导航 系统" 。 重要的是，由于额外的梯度计算是 独立 的，它们可以完全 并行化 ，从而保持 低延迟采样 的特性。 团队引入了一个 两阶段优化框架 ：最初，EPD-Solver通过基于 蒸馏 的方法优化一小组可学习参数；随后，团队进一步提出了一种参数高 效的强化学习微调框架 RDPO ，将求解器重新构建为随机的狄利克雷 （Dirichlet） 策略。 与微调庞大骨干网络的传统方法不同，团队的RL方法严格在 低维求解器空间 内运行，在增强复杂文本到图像 （T2I） 生成任务性能的同 时，有效缓解了奖励作弊 （Reward Hacking） ...

Core Insights - The article discusses the introduction of "Conditional Memory" in Transformer models, which enhances knowledge retrieval mechanisms that were previously lacking in the original architecture [1][2][9]. Group 1: Introduction of Conditional Memory - Conditional Memory is viewed as an essential modeling primitive for the next generation of sparse models [2]. - The research team, led by Liang Wenfeng in collaboration with Peking University, has proposed a new paradigm and implementation plan called the Engram module [3][5]. Group 2: Performance Improvements - The Engram module allows a 27B parameter model to outperform a pure MoE model of the same size, compressing tasks that originally required 6 layers of attention down to 1-2 layers, thus freeing resources for more complex reasoning tasks [5][13]. - The optimal allocation of sparse parameters between MoE and Engram memory results in a U-shaped curve, indicating that allocating about 20% to 25% of sparse parameters to Engram memory minimizes model validation loss [34][36]. Group 3: Technical Implementation - Engram's design incorporates a large vocabulary for static entities and phrases, enabling O(1) speed for information retrieval [7][14]. - The team addresses traditional N-gram model issues, such as semantic redundancy and storage explosion, by compressing tokens and using multiple hash functions to map N-grams to a fixed-size embedding table [22][25]. Group 4: Experimental Results - The Engram-27B model shows significant improvements across various benchmarks, with notable increases in performance metrics such as BBH, ARC-Challenge, and DROP [47]. - The model's architecture allows for efficient memory management, enabling the use of a 100 billion parameter table offloaded to CPU memory without significant latency impact during inference [63][66]. Group 5: Future Developments - The next generation of sparse models from DeepSeek is expected to be released before the Spring Festival, indicating ongoing advancements in AI model architecture [67].

岗位均为全职，工作地点：北京中关村。 编辑部 发自 凹非寺 量子位 | 公众号 QbitAI AI热潮还在汹涌，但如果你还不知道如何参与……那为什么不来 量子位 呢？ 我们是一家以 追踪AI新进展 为核心的内容平台，经过8年积累，目前拥有顶流影响力，广泛且备受认可的产业资源，以及时代风口的最佳观 测和学习生态位。 目前，我们有 三大方向 岗位招聘，希望你是 （或者能成为） 这三个方向的内容专家： 岗位面向： 加入我们，你可以获得： 以下是岗位详情： 所有岗位不同能力层级职位均在开放，欢迎结合个人履历和经验申请。 AI产业方向 岗位职责： AI产业方向 ：关注基建层创新，包含芯片、AI Infra、云计算； AI财经方向 ：关注AI领域创投和财报，跟踪产业链资本动向； AI产品方向 ：关注AI在应用和硬件终端方向的进展。 社招：覆盖编辑、主笔、主编各个层级，按能力匹配岗位； 校招：应届毕业生，接受实习且可转正。 站在AI浪潮之巅 ：第一时间接触和了解AI领域最新技术和产品，构建完整的AI认知体系。 玩转AI新工具 ：将各种AI新技术、新工具应用于工作，提升工作效率和创造力。 打造个人影响力 ：通过撰写独家原创内 ...

Core Insights - The article discusses the emergence of numerous keywords in the AI product sector by 2025, highlighting transformative AI products that are leading the market [4]. - The "AI 100" list by Quantum Bit Think Tank aims to evaluate and recognize the top AI products in China, reflecting both current capabilities and future potential [4][12]. Group 1: AI 100 List Overview - The "AI 100" list is divided into three main categories: "Flagship AI 100," "Innovative AI 100," and the top three products in ten popular sub-sectors [6]. - The "Flagship AI 100" will focus on the strongest AI products of 2025, emphasizing those that demonstrate significant technological breakthroughs and practical value [7]. - The "Innovative AI 100" aims to identify emerging products in 2025 that have the potential to lead industry changes in 2026 [8]. Group 2: Sub-sector Focus - The ten hottest sub-sectors for the top three products include AI browsers, AI agents, AI smart assistants, AI workstations, AI creation, AI education, AI healthcare, AI entertainment, Vibe Coding, and AI consumer hardware [9]. Group 3: Application and Evaluation Criteria - The evaluation of the "AI 100" list employs a dual assessment system combining quantitative and qualitative measures, focusing on user data and expert evaluations [13]. - Quantitative metrics include user scale, growth, activity, and retention, while qualitative assessments consider long-term potential, technology, market space, and user experience [13].

新技术集中火力，重点解决长文本任务的理解、算力难题。 相比于LongCat系列之前的全注意力 MLA机制 ，LoZA只改了一半的核心模块。 但模型长文本能力从256K扩展到1M，解码速度还快了不少。 闻乐 发自 凹非寺 量子位 | 公众号 QbitAI 256K文本预加载提速超50%，还解锁了1M上下文窗口。 美团龙猫 LongCat 系列新年出招，发布 全新稀疏注意力机制LoZA（LongCat ZigZag Attention） 。 甚至比同类型的Qwen-3模型表现还要好。 接下来看具体方案。 如何做到 "只算关键部分" ？ 全注意力机制的算力瓶颈在于平方级的计算复杂度O (L²)，这导致模型在处理长文本任务时对显卡要求高，还会出现推理延迟问题。 LoZA的核心思路是专注于处理重要的内容，不重要的部分少花力气。 作为LongCat系列的核心技术升级，LoZA主要是在原来的MLA机制上做改造。 具体分两步。 首先，给模型里的多头潜在注意力模块MLA做一个全局"筛查"，找出哪些模块可以被改造。 在原来的MLA架构中，每个MLA模块都是处理注意力的核心单元，现在的新方案是给每个模块配一个可学习权重α。 α值越 ...

Core Insights - The article discusses Elon Musk's predictions and insights regarding the future of AI, robotics, and energy, emphasizing the rapid advancements expected in these fields and their implications for society [2][7][30]. Group 1: AI Predictions - Musk predicts that Artificial General Intelligence (AGI) will be achieved by 2026 and that by 2030, AI will surpass the total intelligence of all humans combined [8]. - He believes that current AI has two orders of magnitude of improvement potential, meaning existing hardware could run models that are 100 times smarter [8]. - Musk anticipates a tenfold performance increase in AI capabilities annually, supported by advancements in chip technology and computational power [9]. Group 2: AI Safety - Musk identifies three key traits for ensuring AI safety: truth, curiosity, and beauty [12]. - He argues that truth prevents AI from making irrational decisions, while curiosity ensures that AI values human existence [15]. - The perception of beauty is seen as essential for AI to create a positive future [15]. Group 3: Robotics Advancements - Musk predicts that within three years, Tesla's Optimus robots will surpass the best human surgeons in performing surgeries, with a significant number of these robots deployed [19]. - He explains that the rapid progress in robotics is due to exponential growth in AI software, chip capabilities, and mechanical flexibility [20]. - Musk updates his previous estimate, suggesting that the number of humanoid robots could exceed 10 billion by 2040, with a more immediate increase expected in the next two years [20]. Group 4: Energy and Sustainability - Musk emphasizes the importance of solar energy, stating that humanity currently utilizes only about 1% of the solar energy available on Earth [24]. - He praises China's advancements in solar energy production, predicting that by 2026, China's electricity output will be three times that of the U.S. [26]. - Musk envisions a future where energy becomes the basis of currency, highlighting the potential of space-based data centers powered by solar energy [27]. Group 5: Societal Implications - Musk predicts a future characterized by both high income for all and social unrest, with white-collar jobs being the first to be replaced by AI [32][33]. - He suggests that the transition to an AI-driven economy will be gradual, with companies that fully adopt AI outperforming those that do not [36]. - Musk proposes that the solution to the transition could involve providing everyone with access to goods and services, leading to deflation as production outpaces monetary supply [38].

Core Viewpoint - The article highlights the ongoing momentum in the field of embodied intelligence, particularly focusing on the recent A++ funding round of X Square Robot, which raised 1 billion yuan, indicating strong investor confidence in the company's technology and market potential [2][8]. Funding and Investment - X Square Robot recently completed a 1 billion yuan A++ funding round led by ByteDance and Sequoia China, with participation from top institutions and local platforms [2]. - The company has received investments from major players like Meituan and Alibaba, making it the only embodied intelligence company backed by these three internet giants [3]. - Over the past year, X Square Robot has completed multiple funding rounds, including A+, A, Pre-A++ and Pre-A+++ rounds, showcasing a clear upward trend in financing as technology and products advance [4][5]. - The total funding raised by X Square Robot has exceeded 3 billion yuan across 9 rounds since its establishment, reflecting strong recognition of its independent foundational model technology in embodied intelligence [15]. Technology and Product Development - X Square Robot focuses on self-developed "general embodied intelligence models," with a clear technological path established from the outset [16][17]. - The company has developed the WALL-A series of VLA operational models, integrating perception, understanding, decision-making, and action output into an end-to-end model [20][21]. - The WALL-A model was released in October 2024 and became one of the largest end-to-end unified embodied intelligence models globally, with the WALL-OSS model ranking third in a recent RoboChallenge [22]. - On the hardware side, X Square Robot is advancing two generations of embodied robots, Quantum One and Quantum Two, designed for different operational capabilities [23][25]. Business Strategy - The company adopts a sustainable evolution approach, focusing on building a foundational model that learns in the real physical world, with hardware supporting the model and data feeding back into model iterations [27]. - This strategy has garnered continuous attention and investment from the capital market and industry, indicating a strong belief in the company's long-term vision and capabilities [28].

Core Viewpoint - The article discusses the emergence of KAN (Kolmogorov-Arnold Networks) as a significant advancement in neural network architecture, highlighting its advantages over traditional multi-layer perceptrons (MLPs) in terms of accuracy and interpretability [3][4]. Group 1: KAN Development and Impact - KAN's initial paper was published in April 2024 and quickly gained attention for outperforming MLPs, receiving 1.1k stars on GitHub within a few days [3][4]. - The architecture of KAN offers a new opportunity to improve deep learning models that heavily rely on MLPs, positioning it as a strong alternative [4]. - KAN's design allows for the observation of variable interaction paths, providing interpretability and interactivity that MLPs lack [13]. Group 2: Research Background of Liu Ziming - Liu Ziming, the lead author of KAN, is set to join Tsinghua University as an assistant professor in September 2024, with his first batch of PhD students already recruited [1][7]. - Liu has a strong academic background, having been a top physics student and later pursuing a PhD at MIT under Max Tegmark, focusing on the intersection of physics and machine learning [9][19]. - The inspiration for KAN stems from the Kolmogorov-Arnold theorem, which suggests that complex multi-dimensional functions can be represented as a combination of simpler functions [10][11]. Group 3: Research Philosophy and Future Directions - Liu's research philosophy emphasizes curiosity-driven and impact-driven approaches, aiming for both scientific insight and long-term influence [18]. - He advocates for a combination of theoretical and experimental research, focusing on high-quality abstractions that can be applied across various scientific fields [18]. - Liu maintains a blog titled "physics of AI," where he explores AI phenomena through the lens of physics, aiming to uncover insights that could significantly impact the field [20][24].