Workflow
DeepSeek
icon
Search documents
3700 次预训练寻找 “线性注意力” 非共识,MiniMax-01 开发者讲述 4 年探索
晚点LatePost· 2025-03-09 12:00
"我们跑的是下半场,赌的就是未来的长文本需求。" MiniMax 在今年 1 月发布了参数为 4560 亿的开源大模型 MiniMax-01,该模型就用到了他们开发的线 性注意力机制 "Lightning Attention"。 我们邀请了这个项目的负责人,MiniMax 高级研究总监钟怡然,来与我们一起聊线性注意力的研发过 程。钟怡然在 MiniMax 负责大模型网络架构设计,目前正开发多模态深度推理模型。 钟怡然曾担任上海人工智能实验室青年科学家,是新架构探索组的 PI(项目负责人);他在澳洲国立大 学获得博士学位,师从李宏东教授和 Richard Hartley 院士。他和他的团队已在一些国际顶级学术会议和 期刊上发表了 20 余篇关于模型新架构的论文,覆盖了当前多类非 Transformer 架构,如线性注意力机制 (线性注意力)、长卷积(Long Convolution)和线性循环网络(Linear RNN)。 在 2021 年,线性注意力还是一个 "看起来很美好的泡泡",怡然和团队就开始探索线性架构的实现。 嘉宾 丨 钟怡然 整理 丨 刘倩 程曼祺 上期播客中, 我们与清华的两位博士生,肖朝军和傅 ...
“实习生也月入过万”,AI行业严重缺人
虎嗅APP· 2025-03-09 09:30
Core Viewpoint - The article highlights the intense demand for AI talent in the industry, leading to significant salary increases and a competitive job market for AI-related positions [2][3][4]. Group 1: Salary Trends and Job Opportunities - AI industry salaries are notably high, with DeepSeek offering annual salaries starting at approximately 500,000 yuan, and some positions exceeding 1.76 million yuan [6][7]. - Nearly one-third (30.97%) of top AI job postings have annual salaries above 500,000 yuan, indicating a trend of high compensation in the sector [7][17]. - The demand for AI talent is reflected in the recruitment strategies of major companies like Alibaba and Tencent, which are actively hiring for AI-related roles [23][24]. Group 2: Talent Shortage and Market Dynamics - There is a significant talent shortage in the AI field, with a predicted demand for 6 million skilled AI professionals by 2030, while the supply is expected to be only 2 million, resulting in a 4 million shortfall [22]. - The AI talent shortage is exacerbated by high entry barriers and the need for candidates to possess both technical skills and industry knowledge [27][29]. - Companies are struggling to find qualified candidates, with reports indicating that only 1 in 400 applicants for certain AI roles meet the necessary qualifications [26][27]. Group 3: Future Outlook and Industry Challenges - The article discusses the optimistic outlook for the AI industry, despite potential challenges such as funding requirements and the uncertainty of technology implementation [36][38]. - The competitive landscape among tech giants is leading to resource allocation challenges, as companies develop multiple teams for similar AI projects [37]. - The narrative emphasizes that while there may be some market bubbles, the overall momentum and investment in AI are expected to drive significant advancements in the field [38][39].
两位90后火了
投资界· 2025-03-09 07:47
正在招人,去年估值1亿美金。 作者 I 周佳丽 岳笑笑 报道 I 投资界PEdaily "我们还在尝试接触肖弘团队。"华东一位看大模型的投资人聊起最新一幕,"找他们的人肯定踏破门槛了。" 肖弘 ,1992年生于江西吉安,2 015年从华中科技大学毕业在武汉开始创业,公司卖掉后202 2年启动新的AI项目——Moni c a .im,母 公司叫"蝴蝶效应";公司联合创始人兼首席科学家 季逸超 ,同样是1992年出生,本硕毕业于北京信息科技大学。 Ma nus AI一夜爆火,一种FOMO(害怕错过)情绪也在VC圈隐隐蔓延。据了解,Ma nus已经悄悄完成两轮融资,除了老股东真格基 金之外,还浮现了红杉中国、腾讯以及原美团联合创始人王慧文的身影。 "去年早些时候大概估值1亿美金,现在爆火,已经完全聊不上了。"有投资人心情复杂道。 那是肖弘和他的联合创始人最沮丧的时刻,他们一度决定停止创业,在2016年秋天去北京大厂找到了工作。有了工作o ff e r,肖弘抱 着"玩一下"的心态,参加了当年的黑客松(Ha c ka t hon,一项流行于程序员中的热门活动)。 就是在这场比赛上,他与真格基金合伙人刘元相识。 "第二天 ...
120万年薪!华为小米砸钱抢AI大模型研发人才;我国AI人才缺口达500万人,在校生仅4万人,清华拟扩招150名本科生丨AI周报
创业邦· 2025-03-09 03:27
Core Insights - The article highlights significant developments in the AI industry, focusing on new product launches, funding events, and market trends. Group 1: Product Launches and Innovations - Manus, a general-purpose AI agent from a Chinese team, achieved state-of-the-art performance in the GAIA benchmark, surpassing OpenAI's models, and its invitation code was speculated to be sold for as high as 50,000 yuan [4][5] - Tencent launched and open-sourced a new image-to-video model, allowing users to create 5-second videos from images, available through Tencent Cloud [6] - ByteDance introduced Trae, the first AI-native integrated development environment in China, designed to enhance collaboration between developers and AI [19] Group 2: Market Trends and Competition - The AI investment landscape saw a total of 274.82 billion yuan in disclosed funding events this week, with an average funding amount of 39.26 billion yuan [33] - The competition between Tencent's AI assistant "Tencent Yuanbao" and DeepSeek is characterized as a clash between resource-driven and technology-driven models, with both companies vying for market dominance [6][7] - The global AI talent gap is projected to reach 5 million, with only about 40,000 students currently enrolled in AI-related programs [20] Group 3: Funding and Financial Developments - Anthropic, an AI safety research company, completed a $3.5 billion Series E funding round, emphasizing the growing interest in AI safety and reliability [41] - SoftBank is reportedly negotiating a $16 billion loan specifically for AI investments, indicating a strong commitment to advancing AI projects [28] - OpenAI plans to charge $20,000 per month for its advanced AI agents capable of handling complex tasks, reflecting the increasing monetization of AI technologies [22]
Bold Prediction: 1 Stock That Could Be Worth More Than Nvidia 7 Years From Now
The Motley Fool· 2025-03-08 13:45
Nvidia (NVDA 1.92%) has been one of the best long-term investments of all time. Since 1999, shares have increased in value by more than 285,000%, pushing the company's market capitalization into the trillions of dollars. The cause of Nvidia's soaring valuation has been the rise of artificial intelligence (AI).But Nvidia isn't the only company exposed to the massive tailwind that is AI spending. Long term, there's another chipmaker that could end up giving Nvidia a run for its money. And unlike Nvidia's stoc ...
梁文锋,去香港了?
华尔街见闻· 2025-03-08 09:53
以下文章来源于融中财经 ,作者阿布 融中财经 . 中国领先的股权投资与产业投资媒体平台。聚焦报道中国新经济发展和创新投资全产业链。通过全媒体资讯平台、品牌活动、研究服务、专家咨询、投资顾 问等业务,为政府、企业、投资机构提供一站式专业服务。 作者阿布 编辑吾人 彭博社消息显示,DeepSeek已经于上个月在香港注册了两家公司——DeepSeek Limited 和 DeepSeek (HK) Limited。上述两家公司,在2月5日已经完成注册。 背靠香港,面向全球,香港的站位特点,让其拥有独特的全球化视角。有消息人士认为,DeepSeek此举或有意推动香港总部的建立。 与此同时,香港也着意推进人工智能产业的发展。 就在几周之内,香港首个人工智能大模型正式发布,人工智能安全、可信、负责任论坛在港举办,2025/2026财政年度特区政府财政预算案宣布预留10亿港元成立 香港人工智能研发院…… 连串消息显示,香港正在发力抢进人工智能"新赛道"。 DeepSeek或设立香港总部? "最近见不到梁文锋,他是重点保护'动物'。"一位深圳投资人告诉融中记者。 自从春节复工以来,虽然仅仅只有一个月时间,但梁文锋和DeepS ...
速递|微软“去OpenAI化”计划浮出水面,自研AI模型MAI来了
Z Finance· 2025-03-08 09:44
Core Viewpoint - Microsoft is developing its own AI reasoning models to compete with OpenAI, indicating a strategy to diversify its technology reliance while maintaining its partnership with OpenAI [1][2]. Group 1: Development of AI Models - Microsoft is testing AI models developed by xAI, Meta, and DeepSeek in its Copilot product as potential alternatives to OpenAI's technology [1]. - The AI team, led by Mustafa Suleyman, has completed training a series of models under the codename "MAI," which are nearing the performance levels of top models from OpenAI and Anthropic [2]. Group 2: Strategic Implications - The move to develop independent AI models is part of Microsoft's strategy to reduce reliance on OpenAI's underlying technology and significantly lower operational costs [1][2]. - Microsoft plans to open the MAI models to external developers via an API later this year, allowing integration into their applications [2]. Group 3: Competitive Landscape - The development of chain-of-thought reasoning models by Microsoft's team may lead to direct competition with OpenAI [2]. - The current trend in the AI industry shows tech giants building their own technological moats while maintaining ecosystem partnerships [2].
账号解冻了!Manus最新回应
21世纪经济报道· 2025-03-08 05:03
Core Viewpoint - ManusAI has regained its official account and aims to share innovative use cases, distancing itself from any cryptocurrency-related allegations [1][2]. Group 1: Company Overview - ManusAI is a general-purpose AI agent product that has gained significant attention, being compared to DeepSeek, which recently made headlines [4]. - The company was launched by Beijing Hongse Butterfly Technology Co., Ltd. (Monica.im) on July 3, 2023, and is registered in Haidian District, Beijing [4]. - The founder, Xiao Hong, has a background in software engineering from Huazhong University of Science and Technology and has prior entrepreneurial experience [4]. Group 2: Team and Operations - The co-founder and chief scientist, Ji Yichao, is an active tech entrepreneur, while the product partner, Zhang Tao, has held significant roles in major tech companies [5]. - The company operates in both Beijing and Wuhan, with a total of 60 employees, and has established a branch in Wuhan to handle research and development tasks [5]. - Following the surge in demand for ManusAI's product, teams in both locations coordinated efforts to meet the increased workload [5].
Microsoft reportedly ramps up AI efforts to compete with OpenAI
TechCrunch· 2025-03-07 21:22
Group 1 - Microsoft is intensifying its competition with OpenAI by developing its own AI models and exploring alternatives for products like Copilot [1][2] - The company has created AI "reasoning" models that are comparable to OpenAI's o1 and o3-mini, amid rising tensions due to OpenAI's refusal to share technical details [1] - Microsoft has developed a family of models called MAI, which are competitive with OpenAI's offerings, and is considering providing them through an API later this year [2] Group 2 - Microsoft has invested approximately $14 billion in OpenAI and is diversifying its AI strategy by hiring industry experts like Mustafa Suleyman from DeepMind [3] - The company is testing alternative AI models from xAI, Meta, Anthropic, and DeepSeek as potential replacements for OpenAI technology in its Copilot product [2]
创业邦2025新青年创投榜调研启动
创业邦· 2025-03-07 10:19
Group 1 - The article emphasizes the rise of young entrepreneurs in China, highlighting their role in reshaping the cultural and technological landscape through innovative projects like "Black Myth: Wukong" and "Ne Zha: The Devil's Child" [1] - It discusses the emergence of new technology-driven companies, such as Yushu Technology and DeepSeek, which are pushing the boundaries of AI and robotics, thereby establishing a new narrative in global technology [1] - The article notes that young innovators are becoming a significant driving force in the global innovation economy, with investors playing a crucial role in supporting their visions and translating them into reality [1] Group 2 - The "30 Under 30+" initiative has been launched to identify and celebrate innovative leaders under the age of 35, expanding the previous age limit to include a broader range of young talent [2] - The "40 Under 40" initiative aims to recognize active investors who are pivotal in the venture capital landscape, showcasing successful past winners [2] - The article invites young entrepreneurs and investors to participate in the upcoming 2025 awards, highlighting the importance of recognizing and nurturing new talent in the entrepreneurial ecosystem [2]