第14版的核心是对 QuantLib v1.40 的升级——这是构建 ORE 的开源量化金融库的最新版本。这一集 成带来了更流畅的性能、更高的一致性，并确保ORE能够持续满足全球金融机构不断变化的需求。 此外，自上次发布以来，开发团队已解决了100多个问题单，回应了用户反馈，并对平台进行了微 调，以在所有使用场景下实现更高的稳定性和准确性。 固定收益与大宗商品的扩展建模 本次版本扩展了 ORE 在多类新产品和市场领域的分析能力。新增支持商品期货的美式期权，使得这 些复杂衍生品能够通过有限差分法进行定价。用户现在还可以对可赎回债券及其衍生品进行建模，包 括远期、总回报互换以及基于可赎回债券标的的期权，并能像其他固定收益工具一样实现同样的精确 性和灵活性。 对债券期货的增强功能包括引入"最便宜可交割（Cheapest-to-Deliver）"特性以及债券期货的总回报 互换。同时，为大宗商品浮动利率新增的舍入规则确保估值与市场惯例保持一致。总体而言，这些改 进推动 ORE 不断演进，成为风险管理领域最全面的开源框架之一。 开源技术正在不断重塑金融格局，为企业提供以极低成本甚至免费获取先进分析和模拟工具的途径。 在交 ...

公众号记得加星标⭐️，第一时间看推送不会错过。 据theinformation报道，由Jim Keller领导的芯片公司Tenstorrent在最近一轮融资中估值30亿美金。 Tenstorrent成立于2016年，致力于为NVIDIA的GPU开发可扩展的AI加速器，支持云计算和边缘计 算。该公司目前正在研发RISC-V CPU，并将其设计授权给其他实体。值得一提的是，该公司在其开 发过程中大量采用开源技术，从而避免了使用NVIDIA目前成本高昂的高带宽内存（HBM）。 Tenstorrent 首批上市的芯片是与 GlobalFoundries 合作生产的。Tenstorrent 的 Tensix 处理器由称 为 Tensix Core 的处理器核心组成。每个 Tensix Core 都包含一个用于张量运算的数组数学单元、一 个用于向量运算的 SIMD 单元、一个用于在核心之间和芯片之间传输数据的片上网络 (NoC)、五个 用于辅助控制 NoC 的小型 RISC-V 处理器，以及高达 1.5MB 的 SRAM。 来 源 ： 内容来自半导体行业观察综合 。 Tenstorrent推出的RISC-V替代方案Gra ...

Core Viewpoint - The article highlights the significant achievements of the Southern Power Grid in the open-source domain, particularly in the context of the openEuler community and its contributions to the digital transformation of the power industry [1][2]. Group 1: Event Overview - The openEuler Summit 2025 was successfully held in Beijing, where the Southern Power Grid was recognized for its contributions to open-source practices and awarded titles for leading commercial practices and outstanding contributions [1]. - The summit aimed to release important technical versions of openEuler in the AI field and invited global partners to showcase significant products, marking the beginning of a new chapter for openEuler [1]. Group 2: Technological Innovations - The Southern Power Grid shared its experiences in technology innovation, version planning, practical applications, and ecosystem development during its keynote speech titled "Breakthrough · Innovation · Empowerment" [2]. - The latest version, Electric Hong 3.0, has achieved significant upgrades, supporting hybrid deployment frameworks and enhancing system applicability in core power scenarios [3]. Group 3: Ecosystem Development - The Electric Hong system, developed by the Southern Power Grid, addresses industry challenges by achieving full-stack integration from chips to applications, creating a notable demonstration effect in open-source practices [5]. - Over 500 industry chain manufacturers have joined the Electric Hong ecosystem, with more than 3,000 products included, indicating rapid expansion and collaboration within the ecosystem [5]. Group 4: Future Directions - The Southern Power Grid plans to deepen collaboration with the openEuler and Electric Hong communities, focusing on common industry challenges such as renewable energy consumption and coordinated operations [7]. - By leveraging open-source technology, the company aims to develop comprehensive solutions covering all aspects of the power system, addressing the challenges of achieving a clean, low-carbon, and efficient energy system [7].

Group 1: OpenAI Developments - OpenAI released the GPT-5.1 model, introducing two versions: GPT-5.1 Instant and GPT-5.1 Thinking, with features like adaptive reasoning and improved efficiency [2] - The GPT-5.1 Instant model is designed to be more obedient and reliable in following user instructions, while the Thinking model focuses on efficiency and clarity [2] Group 2: Company Movements - Former DeepSeek researcher Luo Fuli has joined Xiaomi's MiMo team, focusing on building future AGI technologies [3] - Intel's AI executive Sachin Katti is leaving to join OpenAI, indicating a strategic shift in AI infrastructure development [5] - ByteDance terminated an employee for multiple leaks related to its robotics research [6] Group 3: Investment and Financial Activities - AI startup Anthropic plans to invest $50 billion in building data centers in the U.S., creating approximately 800 permanent jobs [4] - Nanwei Semiconductor announced a private placement to raise about $100 million for operational funding and market expansion [13] - ChipLink Capital completed a fundraising of 1.25 billion yuan for its first main fund, focusing on sectors like semiconductors and AI [17] Group 4: Market Trends - Major storage manufacturers, including Samsung and SK Hynix, are planning to increase NAND prices while reducing supply [11] - Jiangbolong reported that demand for large-capacity QLC SSDs is exceeding supply expectations, leading to upward pricing trends [10] Group 5: Corporate Actions - Century Huatong announced a share buyback plan of 500 million to 1 billion yuan, following its stock delisting from risk warnings [19] - Qiangyi Semiconductor's IPO was approved, focusing on the development and production of probe cards for semiconductor testing [20]

Core Insights - Meta has launched the Omnilingual ASR system, capable of recognizing over 1,600 languages, aiming to bridge the digital divide in language technology [1][2][5] - The system supports 500 languages that have never been transcribed by any AI system before, significantly expanding the language coverage compared to existing models like OpenAI's Whisper [5][7] - Omnilingual ASR introduces a flexible learning mechanism that allows the model to learn new languages from minimal examples, potentially expanding its coverage to over 5,400 languages [10][11] Language Coverage and Performance - The system achieves a low character error rate (CER) of less than 10% for 78% of the tested languages, and this rate increases to 95% for languages with over 10 hours of training data [7][8] - It categorizes languages into high-resource, medium-resource, and low-resource, with 95% of high and medium-resource languages achieving a CER below 10% [8] - Even low-resource languages show promise, with 36% achieving a CER below 10% [8] Open Source and Community Engagement - Omnilingual ASR is fully open-sourced on GitHub, allowing researchers, developers, and organizations to use and modify the model without licensing restrictions [11][13] - Meta has released a large multilingual speech dataset, including transcriptions for 350 underrepresented languages, to support community-driven language recognition [13][14] - The development involved collaboration with local language organizations to gather diverse voice samples, ensuring cultural sensitivity and community involvement [15][16] Technical Specifications - The model architecture includes a range of sizes, from lightweight models suitable for mobile devices to larger models with up to 7 billion parameters for high accuracy [16][18] - Training utilized over 4.3 million hours of audio data across 1,239 languages, marking it as one of the largest and most diverse speech training datasets ever [18] - The system's design allows for continuous growth and adaptation, enabling it to evolve alongside the diversity of human languages [18]

Core Insights - The establishment of the first provincial-level open-source community in China marks a significant step towards high-quality development in the open-source ecosystem, with a three-year development plan aiming to incubate over 50 projects and attract 10,000 developers by 2027 [1][2] Group 1 - The signing of a cooperation agreement among Jiangsu Software Industry Talent Development Foundation, Suzhou Industrial and Information Technology Bureau, and Wuzhong District signifies the formal launch of the open-source community project [1] - The global software industry is rapidly evolving towards open-source, platform-based, and ecological models, with open-source technology becoming a core driver of digital economic development [1] - The project aims to overcome challenges in Jiangsu's software industry, such as reliance on key technologies and insufficient ecological collaboration [1] Group 2 - Suzhou Wuzhong District was chosen for its solid industrial foundation, leveraging partnerships with major research institutions and innovation centers to promote the development of the open-source ecosystem [2] - The open-source zone has already gathered over 40 leading enterprises, top universities, and research institutions, currently incubating 15 high-growth open-source projects [2] - The goal is to create a source of open-source technology innovation and an industrial ecological hub that serves Jiangsu, the Yangtze River Delta, and the entire country [2]

Core Insights - The development of AI chips will become easier and cheaper through open-source technology, as stated by chip expert Jim Keller [2] - Keller emphasizes that AI processors are simpler than commonly perceived and that significant investment is not necessarily required for their development [2] - Keller's company, Tenstorrent, is focused on open-source technology for AI processors and compilers, which lowers costs and increases accessibility [2] Group 1: Open-Source Technology - Open-source chips are more cost-effective and accessible, allowing for greater creativity in developing new solutions [2] - Tenstorrent plans to open-source its AI software stack, aiming to create a successful open-source software company [4] - Keller believes that the demand for AI technology is immense and that the next five years will be particularly interesting for the industry [2][4] Group 2: Industry Trends and Future Outlook - Keller predicts that RISC-V architecture will dominate data centers within the next 5 to 10 years, especially in scientific and high-performance computing [4] - The interest from clients in Tenstorrent's IP cores has been described as surprising, indicating a strong market demand [4] - Keller highlights the challenges in the edge AI market, noting that existing IP products are overly concentrated and difficult to program [5] Group 3: Company Background and Leadership - Jim Keller has a notable background, having worked at AMD, Apple, and Tesla, and he became CEO of Tenstorrent in early 2023 [3] - Keller advocates for an open environment where engineers can learn and innovate freely [3] - Tenstorrent recently licensed its Tensix AI accelerator core IP to LG Electronics for use in embedded edge computing applications [4]

Core Viewpoint - LSEG has launched the Risk Analytics Lab, a powerful new platform that combines the flexibility of the Open Risk Engine (ORE) with the efficiency and simplicity of SaaS, designed for risk management professionals to provide smarter, faster, and more collaborative risk management solutions [1][3]. Group 1: Key Features - Risk Analytics Lab offers a free entry point for users to access cutting-edge risk and pricing capabilities, fostering a growing user community for shared innovation and addressing industry challenges [2]. - The platform integrates market data and pre-configured workflows, allowing users to execute risk calculations instantly without lengthy setup processes or complex integration projects [1][4]. - The open-source architecture of Risk Analytics Lab enables users to customize and collaborate, promoting a community-driven approach to pricing and risk analysis [5]. Group 2: Advantages - Risk Analytics Lab provides a free usage option, significantly reducing costs for risk management professionals, making it a more viable investment choice for enterprises [4]. - The SaaS solution can be activated immediately, allowing teams to demonstrate its capabilities out of the box while being scalable to meet complex risk calculation needs [4]. - The platform supports advanced mathematical models and simulations for accurate pricing of structured derivatives, utilizing methods like Black-Scholes and Monte Carlo simulations [7]. Group 3: Applications and Capabilities - Risk Analytics Lab facilitates complex product pricing, xVA simulations, and sensitivity analysis, enabling teams to price complex products and validate models [5][6]. - It integrates with Post Trade Solutions' market data services, providing cleaned market data for pricing and model calibration [7][12]. - The platform supports compliance and regulatory reporting, serving as a technical foundation for Basel III and IFRS 13 reporting [12].

Core Insights - The third Open Atom Competition has been officially launched at the 2025 Open Atom Open Source Ecology Conference, aimed at addressing technical challenges in the open-source field and promoting open-source technology and talent discovery [1][3] - The competition has successfully attracted over 80,000 participants and issued more than 270 million yuan in prizes since its inception in 2023, with a focus on real-world problem-solving and high-value technological outcomes [3] Group 1: Competition Overview - The third Open Atom Competition will continue the open-source philosophy of "co-construction, co-governance, and sharing," focusing on key areas such as basic software, industrial software, and artificial intelligence, with around 50 competition items planned [3] - The competition is open to individuals, teams, enterprises, and universities globally, providing official certificates and substantial prizes to winners, as well as internship and interview opportunities with major companies like Huawei, Tencent, Inspur, and vivo [3] Group 2: AI Focus and Technical Challenges - The recent roadshow highlighted AI technology innovation and open-source ecosystem development, featuring multiple significant competition items in cutting-edge fields such as operating systems, large models, and robotics [3][4] - Specific challenges include the openKylin series, which offers four tasks with a total prize of 1 million yuan, focusing on optimizing openKylin's intelligent engine and developing multi-agent interaction technology [3][4] Group 3: Developer Support and Engagement - The competition organizers have established dedicated communication communities to enhance participant experience, offering technical documentation, development tools, and hardware support throughout the competition cycle [5] - Various industry experts shared insights during the event, showcasing technological breakthroughs and applications of open-source models in real-world scenarios, further encouraging developer participation [5]

Core Viewpoint - The "Open for SDG Conference" aims to accelerate the achievement of the United Nations Sustainable Development Goals (SDGs) through open collaboration and innovation in artificial intelligence (AI) and open-source technology, scheduled for September 13, 2025, in Hangzhou, China [1][5][7]. Group 1: Conference Objectives and Themes - The conference focuses on exploring how AI and open-source technologies can address global challenges such as poverty, hunger, health, education, and climate change [1][7]. - It serves as a key platform for the UN's "AI for SDGs" initiative, discussing four main themes: Open Data, Open Policy, Open Science, and Future AI, emphasizing the integration of technology with sustainable development goals [7][39]. Group 2: Notable Speakers and Participants - The conference will feature influential speakers, including Huang Tiejun, a professor at Peking University, and Mehdi Snene, Chief AI Officer at the UN, who will set the tone for the event [10][11]. - A diverse array of experts from academia, industry, and the open-source community will share insights on how AI and open-source can empower sustainable development [9][45]. Group 3: Action-Oriented Initiatives - The conference will include a "Show & Tell" segment to showcase practical examples and open-source projects contributing to the SDGs, transforming dialogue into actionable inspiration [45]. - Initiatives such as an open contribution framework and an open data initiative will be launched to facilitate meaningful participation in open-source projects aligned with the SDGs [46][47]. Group 4: Future Vision and Collaboration - The conference aims to catalyze global collaboration among innovators from various fields to identify and undertake impactful projects supporting the SDGs, with a progress report to be submitted at the UN OSPOS for Good conference in 2026 [47][48]. - It emphasizes the importance of creating partnerships and translating shared visions into a more equitable, prosperous, and sustainable future [48].