Core Viewpoint - DeepSeek has launched DeepSeek-Prover-V2-671B, marking a significant advancement in AI mathematical reasoning capabilities, particularly in automated theorem proving [2][4]. Group 1: Model Overview - DeepSeek-Prover-V2-671B is a next-generation automated theorem proving expert model with 671 billion parameters, optimized for proof generation and verification in the Lean 4 framework [4][6]. - The model employs a mixture of experts (MoE) architecture, activating approximately 37 billion parameters per inference, enhancing computational efficiency while maintaining strong reasoning capabilities [4][6]. Group 2: Key Breakthroughs - The release signifies three major milestones, including the potential for innovation across various application domains [6]. - The model's specifications include a context length of approximately 128,000 tokens, allowing it to handle complex reasoning chains and lengthy proofs [6][7]. - The attention mechanism is likely a multi-head latent attention (MLA), which compresses key-value (KV) cache, significantly reducing memory requirements [6][7]. Group 3: Applications and Impact - The model supports formal verification in areas such as cryptographic security proofs and chip design validation, enabling rigorous mathematical checks in automated processes [7]. - It aids mathematicians in formalizing theorems, exploring new conjectures, and proving complex mathematical problems, potentially accelerating mathematical research [7]. - The model can be utilized as an interactive educational tool, guiding students in mastering rigorous mathematical proof methods [7].

| 超微电脑 | V | | | | | --- | --- | --- | --- | --- | | SMCI 已收盘 04-30 16:00:00 美东 | | | | | | 2.94万人加自选(一 | | | | | | 31.86 -4.14 -11.50% | US 齡 空 期 LO | | | | | 高 32.00 | 总市值 190.14亿 。 | 开 29.12 量 9823.05万股 | | | | 市盈TTM 13.16 | 低 28.78 | 换 16.46% | 额 29.81亿 | | | 期权 成交量71.31万张 未平仓数250.76万张 | | | | | | 盘后 31.95 +0.09 +0.28% | 19:59:57 美东时间 | | | | | 分时 | 五日 日K | 李K 年K 分钟, | | | | 均价:30.37 最新:31.86 -4.14 -11.50% | 0.00% | 36.00 | | | | 15:59 31.86 | 100 | | | | | 15:59 31.85 | 300 | | | | | 100 | 15:59 31.86 ...

完整早报音频，请点击标题下方小耳机收听 【苹果重组全球事务和音乐部门】 据知情人士透露，苹果公司正在对其全球事务和音乐部门的管理层分别进行改组，延续了这家iPhone生产商最近的一系 列变动。上述人士说，此次全球事务重组包括调整欧洲、印度、中国和亚洲其他地区政府团队的管理。由于人事变动尚未公布，这些人士要求不具名。与 此同时，Apple Music将有一个全新的领导结构——两名联席主管向奥利弗·舒瑟（Oliver Schusser）汇报工作。舒瑟是苹果公司的高级副总裁，曾领导过该 部门。（财联社） 【OpenAI回应GPT-4o更新后个性过于谄媚：已回滚到老版本】 OpenAI首席执行官山姆·奥特曼在社交平台表示，昨晚开始回滚GPT-4o的最新更新，现在 免费版的回滚已100%完成，付费版完成后会再次进行更新，预计晚些时候对模型个性进行额外的修复，并将在未来几天分享更多信息。此前，奥特曼发文 称，"GPT-4o的最近几次更新使其个性变得过于谄媚和烦人（尽管其中也有一些非常好的部分），我们正在尽快修复。"（搜狐） | Sam Altman > @ @sama · 6小时 | | | | | --- | --- ...

Core Insights - DeepSeek has officially released DeepSeek-Prover-V2 on Hugging Face, continuing its open-source momentum with two versions launched [1][4] - The training core of DeepSeek-Prover-V2 combines "recursion + reinforcement learning," enabling the model to break down complex theorems into sub-goals and reasoning paths [3][8] Model Specifications - DeepSeek-Prover-V2-7B is based on the previous V1.5 model and supports a maximum context input of 32K [4] - DeepSeek-Prover-V2-671B is trained on the DeepSeek-V3-Base, showcasing the strongest reasoning performance [4] Training Process - The training process consists of two phases: the first phase focuses on rapid mode using an "expert iteration" method, where successful answers refine the model [5] - In the second phase, more complex logical reasoning capabilities are trained, incorporating mathematical knowledge from DeepSeek-V3 and formal data [6] Reinforcement Learning - The GRPO reinforcement learning algorithm is introduced to enhance reasoning capabilities, allowing the model to autonomously learn to select optimal solutions from multiple candidates [8] - The system generates 32 different proof schemes for each theorem, retaining only those verified as correct by the Lean verification system [9] Model Distillation - After developing the powerful 671B model, the team distilled its capabilities into a smaller 7B model, allowing users to achieve near-equivalent mathematical reasoning abilities on resource-limited devices [10][11] Reasoning Modes - The rapid mode (non-CoT) focuses on speed, generating concise Lean code answers without showing the thought process, suitable for handling numerous problems [12] - The logical mode (CoT) details each step of the reasoning process, ensuring clarity and transparency [12] Performance Evaluation - In the final performance assessment, DeepSeek-Prover-V2-671B achieved an 88.9% pass rate in the MiniF2F test, successfully solving 49 problems from the PutnamBench dataset [17] New Dataset - DeepSeek introduced a new formal mathematical dataset, ProverBench, containing 325 problems across various mathematical domains, including number theory, algebra, and calculus [18][19] Comparison and Trends - The comparison shows a significant trend: the performance gap between large language models in "informal mathematical reasoning" and "formal mathematical reasoning" is narrowing [21] - The evolution of model structure and training strategies enables models to produce rigorous, verifiable mathematical proofs [22] Future Directions - DeepSeek-Prover-V2 indicates a shift in focus from merely generating content to generating structured logic, which may touch upon the foundational structure of general artificial intelligence [33][34]

中国电子 China (Overseas) Technology 国产开源模型千帆竞发，阿里 Qwen-3、小米 MiMo、DeepSeek Prover 集中发布 A surge of domestic open-source models 姚书桥 Barney Yao 吴叡霖 Louis Ng wo[Table_Title] Research Report 30 Apr 2025 barney.sq.yao@htisec.com louis.yl.ng@htisec.com [Table_yemei1] Flash Analysis [Table_summary] 事件： 2025 年 4 月 28 日，阿里正式发布了新一代 Qwen-3 系列大语言模型（LLMs），包括从百亿参数到数十亿参数多个 量级的模型版本。2025 年 4 月 30 日，小米正式发布并开源了其首个专为推理任务设计的大语言模型——Xiaomi MiMo；DeepSeek 团队在 Hugging Face 平台发布了其最新的大语言模型——DeepSeek-Prover-V2-671B 点评： 阿里 Qwen-3： Qwen-3 系列具 ...

Domestic News - The manufacturing Purchasing Managers' Index (PMI) for April is reported at 49.0%, a decrease of 1.5 percentage points from the previous month, indicating a decline in manufacturing activity [3] - The new Private Economy Promotion Law will come into effect on May 20, aiming to support the development of the private sector [4] - The total holdings of gold ETFs in the Chinese market have reached a historical high, as reported by the World Gold Council [6] - The People's Bank of China conducted a 12 billion yuan reverse repurchase operation using a fixed quantity and interest rate bidding method [11] International News - Traders are fully pricing in four rate cuts of 25 basis points by the Federal Reserve by the end of 2025 [1] - Global gold demand in Q1 reached the highest level for a first quarter since 2016, according to the World Gold Council [2] - The U.S. economy has contracted, with a reported GDP decline of 0.3% in the first quarter, marking the first economic shrinkage since 2022 [6]

就在所有人都在期待DeepSeek官宣R2大模型之际，公司却出其不意地在"五一"前夕投下了另一枚技术炸弹。 4月30日，DeepSeek在Hugging Face平台上悄然开源了其最新模型——DeepSeek-Prover-V2-671B，一个专注于数学定理证明的大语言模型，专门针 对形式化数学证明任务进行优化。 DeepSeek-Prover-V2-671B使用了DeepSeek-V3架构，参数高达6710亿，采用MoE（混合专家）模式，具有61层Transformer层，7168维隐藏层。 | Hugging Face Q. Search models, datasets, users ... | | Models | ■ Datasets ■ Spaces Posts | Docs | Enterprise | Pricing | VII | Log In Sign Up | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | < deepseek-ai/DeepSeek-Prover-V2-671B = 0 Wke 152 | Follo ...

本文来自微信公众号： 机器之心 ，作者：杨文、Panda，题图来自：AI生成 昨晚，奥特曼在 X 上发了条帖子，大意是由于发现 GPT-4o "过于谄媚"的问题，所以从周一晚上开始回滚 GPT-4o 的最新更新。 免费 ChatGPT 用户已 100% 回滚，付费用户完成回滚后会再次更新。同时，他还透露，团队正在对模型个性进行额外的修复，并将在未来几天分享更 多信息。 就在刚刚，OpenAI 还专门发博客来回应此事，详细解释了事情的经过以及他们如何处理模型"拍马屁"的情况。 OpenAI 也指出，这个问题很重要。ChatGPT"阿谀奉承"的性格影响了大家对它的信任和使用体验。如果它总是说好听、但不真诚的话，就会让人觉得 它不可靠，甚至有些烦。 为了解决大模型过度逢迎的问题，OpenAI 除了撤销最新的 GPT-4o 更新外，还采取了更多措施： 目前，用户可以通过自定义指令等功能，给模型提供具体指示来塑造其行为。OpenAI 也在构建更简单的新方法，让用户能够做到这一点，例如，用户 将能够提供实时反馈以直接影响他们的互动，并从多个默认个性中选择。 优化核心训练技术与系统提示：明确引导模型避免阿谀奉承。 增加更多 ...

近日，Meta首席执行官马克·扎克伯格接受了媒体采访，全程信息量满满。访谈中， 扎克伯格谈到了 Meta如何看待下一步AI发展格局，并回应了外界认 为"DeepSeek吊打Meta"的质疑。 他表示，通过比较Llama 4 模型与 DeepSeek 的能力可知， 尽管 DeepSeek 可能在特定领域取得了显著进展，但Llama 4模型能够提供更高的效率和更广泛 的功能。 以下为采访内容（有删节）： 马克·扎克伯格：在我看来，世界会变得更加有趣、甚至有些奇特。根据我的经验，如果你觉得别人做的事情不好，但他们自己却认为很有价值，那么通 常是他们对，你错了。 主持人Patel： 我们似乎正在消除技术利用奖励机制来完全操纵我们的所有障碍。 马克·扎克伯格：我们正在努力构建能推进 Llama 研究的编码代理。我估计 在未来 12 到 18 个月内，我们将达到一个阶段，届时这些研发工作所需的大部 分代码都将由 AI 编写。我倾向于认为，至少在可预见的未来，这反而会增加对人类工作的需求，而非减少。如果你将提供服务的成本降至原来的十分之 一，那么现在去做这件事实际上可能是有意义的。 主持人Patel：你上次来的时候，发布了 ...

温馨提示 ： 点击下方图片，查看运营团队2025年最新原创报告（共210页） 说明： 欢迎约稿、刊例合作、行业人士交流 ， 行业交流记得先加入 "机器人头条"知识星球 ，后添加（ 微信号：lietou100w ） 微信； 若有侵权、改稿请联系编辑运营（微信：li_sir_2020）； 人形机器人要实现真正商业化落地是上肢重要还是下肢重要？ 人形机器人真正落地实用场景，任务终结点是手臂和手，而小编注意到涉及手臂相关研究极少，是工业机械臂发 展多年，导致人形机器人机械臂结构和相关算法控制已完全成熟，只需要专注于"小脑"上层层面控制？但是参加 展会时，可明显看到人形机器人手臂运动过程中颤颤巍巍、卡顿、僵硬的现状，所 以这是"小脑"层面控制问题， 还是关节间问题？ 目前业内对于灵巧手研究已经很多且备受重视，除了人形机器人本体厂商自研外，也出现了专注于灵巧手和触觉 感知研究的初创公司，也是目前人形机器人发展过程中核心卡点之一。 正文： 具身智能机器人是一个复杂的AI+机器人+自动驾驶的系统性学术+工程问题，远期AGI的物理世界载体，受算力、 软件算法、数据、硬件、工程化等多面因素影响；小编往 期文章 ： 【原创】人形机 ...