腾讯QQ已完成鸿蒙电脑适配 微信等将陆续上线 《科创板日报》19日讯，《科创板日报》从业内知情人士处获悉，腾讯QQ已完成鸿蒙电脑适配，微 信、企业微信正在加快体验适配，并将陆续上线。（记者 黄心怡） ...

Core Viewpoint - Tencent has launched the latest Mixed Yuan Image 2.0 model, which claims to revolutionize the traditional "draw card - wait - draw card" method by achieving real-time image generation, enhancing interactive experiences in the industry [1]. Group 1: Model Features - The Mixed Yuan Image 2.0 model emphasizes speed, supporting both text-to-image and drawing-to-image generation, allowing users to receive high-quality images in milliseconds regardless of input method [1][4]. - The model allows for real-time modifications on images using a drawing board, significantly improving efficiency compared to traditional AI image generation methods [4][7]. - Compared to its predecessor, the model's parameter count has increased by an order of magnitude, benefiting from a highly compressed image codec and a new diffusion architecture, resulting in faster image generation speeds [7]. Group 2: Performance Metrics - In a benchmark evaluation (GenEval), the Mixed Yuan Image 2.0 model achieved an accuracy rate exceeding 95%, outperforming other similar models in understanding and generating complex text instructions [8]. - The model's performance metrics indicate it leads in various categories, such as single object and two object generation, with a score of 0.9597 in overall image generation [8]. Group 3: User Experience - Demonstration cases show that users can input commands and see immediate changes in the generated images, enhancing the creative process and allowing for quick adjustments [3][5]. - The model's ability to generate images while users continue to input commands represents a significant advancement in user interaction and experience [7].

Core Viewpoint - Tencent has launched the Hunyuan Image 2.0 model, which features real-time image generation and significantly improved image quality and interaction experience compared to its predecessor [1][3]. Group 1: Model Performance - The Hunyuan Image 2.0 model has increased its parameter count by an order of magnitude, utilizing a high-compression image codec and a new diffusion architecture, achieving millisecond-level response times for image generation [3]. - The model's image generation quality has improved, effectively avoiding the "AI flavor" commonly found in AIGC images, resulting in high realism and rich details [3][4]. - In the GenEval benchmark for complex text instruction understanding and generation, the model achieved an accuracy rate exceeding 95%, outperforming other similar models [4]. Group 2: User Experience - The model allows users to generate images while typing or speaking, transforming the traditional "draw-wait-draw" process into a more interactive experience [3][6]. - A real-time drawing board feature has been introduced, enabling users to see coloring effects as they sketch or adjust parameters, enhancing the creative process for professional designers [13]. Group 3: Future Developments - Tencent hinted at the upcoming release of a native multimodal image generation model, which will excel in multi-round image generation and real-time interaction [15].

分享嘉宾 | 郭春超 责编 | 梦依丹 出品丨AI 科技大本营（ID：rgznai100） 现在这个时代，我们玩的游戏、看的电影、甚至未来的虚拟世界，都离不开精细逼真的三维（3D）模型。然而，制作这些 3D 内容，过去常常意味着 耗时数周甚至数月的人工建模，成本高昂且效率低下。就像平面设计曾被 Photoshop 改变一样，人工智能正在瞄准 3D 领域，试图彻底革新数字内容 的生产方式。 在这场由 AI 驱动的 3D 生成浪潮中，腾讯混元团队推出的开源项目 Hunyuan 3D 成为了全球开发者社区的焦点。它不仅在 GitHub 上迅速积累了超过 9.6k 的 Star，跻身 3D 生成开源项目的第一梯队，更凭借其出色的模型生成效果，赢得了"几乎没有变形的 Image to 3D，恐怖如斯"这样的用户评 价。 AI 生成 3D 的能力发展到什么阶段了？它离真正改变游戏、影视、数字人等行业还有多远？ 在 4 月 18-19 日举行的 2025 全球机器学习技术大会 （ML-Summit）上，腾讯混元 3D 负责人 郭春超 对此进行了详尽解读，并在会后接受了 CSDN 专访。 令人意外的是，尽管当前 3D AIG ...

Core Viewpoint - The article emphasizes the transformative impact of AI technology on the job market for programmers, highlighting the urgent need for professionals to adapt to new AI-driven roles and skills to avoid career obsolescence [1]. Group 1: Industry Trends - Major tech companies like Alibaba Cloud, ByteDance, Tencent, JD.com, and Baidu are integrating AI capabilities into their core operations, with a significant portion of job openings (80%) related to AI [1]. - Traditional programming skills are becoming less relevant, as the demand shifts towards expertise in AI model development and application [1]. - The job market is experiencing a dichotomy where traditional roles are shrinking while AI-related positions are expanding, with salaries for AI roles increasing by 150% [1]. Group 2: Career Development - Professionals are encouraged to acquire skills in AI model principles, application technologies, and practical project experience to remain competitive in the evolving job landscape [1][10]. - The article promotes a free training program aimed at equipping participants with essential AI model development skills, emphasizing the importance of early engagement in AI education [2][3]. - The course offers insights into typical business scenarios, technical architecture, and knowledge systems related to AI model applications, aiming to enhance participants' career prospects [3][5]. Group 3: Course Offerings - The training program includes live sessions and interactive Q&A with industry experts, focusing on practical applications of AI technologies like RAG and Fine-tuning [5][12]. - Participants will receive a collection of AI model application case studies and a white paper on AI commercialization upon completion of the course [12][16]. - The course aims to provide a comprehensive understanding of AI model principles and real-world applications, helping participants build a competitive edge in the job market [13][15].

Group 1 - Microsoft plans to cut less than 3% of its global workforce, approximately 6,000 employees, with notifications starting May 13 [1] - Affected employees will remain on the payroll for 60 days and will still be eligible for rewards and bonuses [1] - The cuts aim to reduce the number of middle managers and increase the ratio of coders to non-coders on projects [2] Group 2 - The tech industry is experiencing a trend of reducing middle management, with Amazon and Google also making similar cuts [3] - Microsoft is focusing on decreasing the "PM ratio," which refers to the ratio of product managers or program managers to engineers [3]

Group 1 - The core viewpoint of the article highlights advancements in AI technology across various companies, focusing on their applications in robotics, e-commerce, and fraud prevention [1][2][3][4]. Group 2 - Songyan Power has launched a new generation of female bionic robot named "Xiao Nuo," featuring multi-modal interaction, 32 degrees of freedom (DOF), and near-human facial expressions, with over 1,000 units already pre-ordered [1]. - OpenAI and Microsoft are reportedly in discussions to unlock new financing and explore future IPO opportunities [2]. - Alibaba's chairman, Cai Chongxin, emphasized the company's focus on two core strategic directions: e-commerce and "Cloud + AI," aiming to integrate AI into all business segments over the next three to five years [2]. - Google has introduced AI-driven fraud detection systems in its applications and search engine, claiming to block "hundreds of millions" of fraudulent search results daily, with a 20-fold increase in efficiency compared to three years ago [3][4]. - Google Messages and Phone applications now feature AI capabilities to detect potential fraud, alerting users to reduce the risk of phone scams [4].

Core Insights - Tencent's technical team has optimized the DeepEP communication framework, achieving significant performance improvements in various network environments, with a 100% enhancement in RoCE and a 30% enhancement in IB networks, facilitating more efficient AI large model training solutions [2][3] - The optimization addresses key bottlenecks in the original DeepEP framework, particularly in bandwidth utilization and CPU control delays, which were limiting its broader application [2][3] Group 1 - The optimization includes intelligent bandwidth allocation through topology-aware multi-QP chaining technology, ensuring full utilization of dual-port network card bandwidth and preventing bandwidth waste [3] - Tencent has resolved CPU control bottlenecks in GPU communication by optimizing the control plane operations to bypass CPU intermediaries, reducing latency and energy consumption [3] - A new "QP internal sequencing lock" mechanism has been introduced to ensure accurate and sequential data transmission among multiple GPUs, even when handling over 1,000 simultaneous data transfer tasks [3] Group 2 - The optimized DeepEP framework has been fully open-sourced and successfully applied in Tencent's mixed Yuan large model training and inference projects, demonstrating excellent versatility in high-performance environments built with Tencent's Xingmai and H20 servers [3]

Core Viewpoint - The article emphasizes that AI technology is significantly transforming the job landscape for programmers, with a shift from traditional development roles to positions requiring expertise in AI and large models [1][12]. Group 1: Industry Trends - Major companies like Alibaba Cloud, ByteDance, Tencent, JD.com, and Baidu are integrating AI into their core operations, with a notable increase in job openings related to AI [1]. - Traditional development frameworks and big data tools are losing their relevance, as businesses prioritize skills in AI model application and development [1][12]. - The demand for AI-related technical positions is surging, with salaries increasing by 150%, and top companies offering annual salaries between 700,000 to 1,000,000 yuan for AI talent [1][10]. Group 2: Career Development - The article highlights the urgency for programmers to acquire skills in AI model principles, application technologies, and practical project experience to remain competitive in the job market [1][12]. - A free training program titled "Large Model Application Development Practical Training Camp" is being offered to help individuals enhance their core competencies in AI [2][3]. - The program includes live sessions covering career prospects, typical business scenarios, technical architecture, and hands-on experience with AI technologies like Transformer architecture and Fine-tuning [3][7]. Group 3: Learning Opportunities - The training camp aims to equip participants with the necessary skills to navigate the evolving job market and avoid potential career pitfalls, especially for those nearing the age of 35 [12][19]. - Participants will receive practical resources, including a collection of large model application cases and a white paper on AI commercialization, to aid in their learning and job applications [15][19]. - The course promises insights from industry leaders and opportunities for direct referrals to major companies, enhancing participants' chances of securing high-paying positions [17][19].

当王女士点开手机里那段AI生成的动态影像时,屏幕上的母亲仿佛重新活了过来——她侧过身来,嘴角泛 起熟悉的笑容,张开双臂将年幼的女儿拥入怀中。这段短短5秒的视频,由一张20多年前的老照片转化而 来。自母亲离世以来,王女士再度感受到"被母亲拥抱"的真实触感:"妈妈又一次抱了我,好像她还在我身 边。" 一刻相册的破局之道,源于对技术温度的深刻理解。其自主研发的FaceFlow算法,将单帧静态图像的3D姿 态重建精度提升至行业新高度——通过解析人物骨骼运动逻辑与面部微表情肌群关联,即使是泛黄的老 照片,也能"学会"自然眨眼、微笑等微表情,乃至做出拥抱等复杂动作。 技术团队透露,这项针对亚洲人优化的算法实现了98.7%的脸型适配率,较开源模型提升70%。当用户上传 照片后,系统首先进行画质修复,接着重建3D人脸模型,最终生成16fps的影视级流畅动图。对比传统GIF的 简单位移变形,这种基于人体运动学的模拟,能还原发丝飘动时的空气阻力效应,甚至保留母亲怀抱孩子时 衣褶的细微波动。 "技术不该是冷冰冰的代码。"一刻相册一位产品负责人强调。为了让更多用户跨越技术鸿沟,一刻相册 将复杂算法封装为"抱抱"模板,只需要在APP内 ...