Core Viewpoint - The M-Robots open-source project has been officially launched, aiming to create a unified robot operating system based on open-source HarmonyOS, promoting ecosystem integration, capability reuse, and intelligent collaboration in the robotics industry [1] Group 1 - The project is incubated by the Open Atom Open Source Foundation and led by Deep Open Hong [1] - The initiative focuses on collaborative development through open-source methods [1] - The goal is to enhance the robotics industry's ecosystem and foster intelligent cooperation among various robotic systems [1]

You really look at the way in which we can implement AI from a governance perspective, from a risk tolerance perspective, from a cultural perspective within businesses and arena. When you're looking at this report, when you're thinking about AI being more hands off, more pro-growth, is that the right way. Well, good to see you, Caroline.And today, we really commend the White House on launching the action plan. I think if you dive a little bit deeper into the action plan, there's a key, key statement and sen ...

Open source X rival Mastodon begins raising funds with new in-app donation feature | TechCrunch https://t.co/grPtBLwNLX ...

Core Viewpoint - The article highlights the launch of Qwen3-Coder by Alibaba's Tongyi team, which has set a new state-of-the-art (SOTA) in AI programming, surpassing both open-source and closed-source models in the industry [1][3]. Group 1: Product Features - Qwen3-Coder includes multiple versions, with the strongest being Qwen3-Coder-480B-A35B-Instruct, featuring a 450 billion MoE model and 35 billion active parameters [5]. - It natively supports a context length of 256K and can be extended to 1 million using YaRN technology [6][23]. - The command-line version, Qwen Code, has been developed based on Gemini Code, allowing for prompt and tool invocation protocol adaptation [8]. Group 2: Performance and Capabilities - Users have reported impressive results, such as creating interactive animations and dynamic weather cards with simple prompts [11][13]. - The model can easily generate a playable Minesweeper game and an editable resume template, showcasing its versatility [16][19]. Group 3: Technical Details - During the pre-training phase, Qwen3-Coder utilized various scaling techniques to enhance model capabilities, with training data totaling 7.5 trillion tokens, 70% of which is code data [22]. - The post-training phase involved scaling code reinforcement learning (RL) to improve the model's performance on real-world coding tasks, achieving high success rates in code execution [24][27]. Group 4: Open Source vs Closed Source - Qwen3-Coder is open-source under the Apache License Version 2.0, making it commercially friendly and empowering developers [29][30]. - The article emphasizes that this release represents a significant leap for open-source programming agents, positioning Chinese models at the forefront of the industry [34].

HEDERA is major open sourceHedera (@hedera):Hedera is fully open source 🛠️ Built by the community, shaped by devs, node runners, and ecosystem builders.Hiero, part of @lfdecentralized, is the open-source codebase that powers Hedera 🤝 Anyone can read it, contribute, and build with it.Want changes to the network? https://t.co/wKLs3lHjVg ...

Core Viewpoint - Perplexity has launched a new web browser named Comet, aiming to challenge Google Chrome's dominance in the market, which currently holds a 66.6% market share. The launch coincides with rumors of OpenAI's own browser release, indicating a competitive landscape in AI-driven search and browsing tools [1][2][3]. Group 1: Product Launch and Market Strategy - Comet integrates Perplexity's AI search tools and smart assistant to enhance user experience, initially available to premium users at $200 per month [1]. - Perplexity's ambition extends beyond user acquisition; they aim to replicate and potentially surpass Google's business model [1][2]. - The company has expressed willingness to acquire Google Chrome if legal pressures force Google to divest it, indicating a strategic move to capture a larger market share [1]. Group 2: Data Acquisition and Advertising - CEO Aravind Srinivas highlighted the importance of gathering user behavior data outside of Perplexity's applications to improve advertising quality, framing the browser as part of a broader data strategy [2]. - The decision to create a browser stemmed from a rejection by Google to include Perplexity as a default search engine, prompting the company to develop its own solution [2][3]. Group 3: Competitive Landscape and Industry Commentary - The Comet browser is built on Google's open-source Chromium project, which raises questions about the originality of Perplexity's offering [3]. - The current trend in startups involves forking open-source projects to add paid features, reflecting a broader entrepreneurial strategy in the tech industry [4]. Group 4: Vision and Long-term Goals - Perplexity aims to leverage AI to transform decision-making processes in finance, targeting a market valued at tens of trillions of dollars, significantly larger than Google's annual revenue [8][36]. - The company seeks to disrupt the financial research market dominated by Bloomberg, proposing that AI can enhance decision-making efficiency and democratize access to financial insights [36][37]. Group 5: Product Features and User Experience - Comet is envisioned as a "cognitive operating system" that integrates AI into daily workflows, allowing users to automate tasks and improve productivity [14][15]. - The browser will enable users to issue commands directly, with AI handling tasks such as data extraction and report generation, enhancing the overall user experience [15][34]. Group 6: Funding and Investor Relations - Perplexity's development costs were notably low, with the product being built for just $50,000, which impressed investors like Marc Andreessen [7][28]. - The company has faced skepticism from investors who prefer safer, vertical market strategies, but Srinivas remains committed to tackling larger, more challenging problems [6][26].

Finally, a framework to connect any LLM to any MCP server (open-source).mcp-use lets you connect any LLM to any MCP server & build custom MCP Agents, without using closed-source apps like Cursor/Claude.Compatible with Ollama, LangChain, etc.Build 100% local MCP clients! https://t.co/8rhqh7BUZh ...

Core Viewpoints - The article highlights significant developments in the AI sector, including investment activities and technological advancements, providing insights into the global AI market trends during the week of July 12-18 [1]. Domestic Major Events - Yushu Technology has initiated its listing guidance with CITIC Securities as the advisory firm, with the controlling shareholder holding 34.76% of the company [3]. - ByteDance's visual model leader, Yang Jianchao, announced a temporary break, with Zhou Chang taking over his responsibilities [3]. - NVIDIA's CEO Jensen Huang emphasized the importance of open-source models in benefiting various industries, including healthcare and finance, during a conference [3][6]. - Yushu Technology has published a patent for a laser radar technology that enhances angle tracking precision and eliminates jitter interference [4]. Investment and Financing Overview - MiniMax is set to complete nearly $300 million in new financing, raising its valuation to over $4 billion, and is seeking an A-share listing [4]. - The week saw a total of 25 disclosed AI financing events globally, a decrease of 20 from the previous week, with a total financing scale of 5.375 billion RMB [43]. - In the domestic market, MiniMax led with a financing amount of nearly $300 million, focusing on multi-modal AI model development [52]. International Major Events - CoreWeave plans to invest $6 billion to establish an AI data center in Pennsylvania, with an initial capacity of 100 megawatts [28]. - OpenAI launched the ChatGPT Agent, capable of autonomous thinking and action [29]. - Meta has completed the acquisition of voice AI startup PlayAI, integrating its team into Meta [14]. - Apple is considering acquiring European AI startup Mistral, which has raised €1.1 billion, potentially marking its largest acquisition [34]. AI Financing Events - The highest disclosed financing in the overseas AI sector was for medical AI service provider OpenEvidence, which completed a $210 million Series B financing [55]. - The domestic AI financing events were primarily concentrated in Beijing and Jiangsu, with each reporting six events [48].

Open Thoughts项目概览 - Bespoke Labs 发布 Open Thoughts 3，旨在创建最佳的开源推理数据集 [1][9] - Open Thoughts 项目专注于推理数据配方，以解决创建强大推理模型的关键缺失环节 [6][9] - Open Thoughts 3 在科学、代码和数学等领域都优于 Deepseek R1 quen 7B 模型 [13] 数据集创建与优化 - 数据集流水线包括问题来源、混合、过滤、答案生成和答案过滤等步骤 [17] - 实验创建了超过 5000 个数据集和近 3000 个模型，以严格评估流水线中每个步骤的不同决策 [18] - 每个问题采样多个推理轨迹效果显著，在固定问题规模下，性能不会下降，允许数据规模扩大 16 倍 [19][20] - 合成问题是可扩展的，可以进一步提高准确性 [22] - 问题过滤通过让语言模型评估问题的难度和答案的长度来筛选高质量问题 [23] 关键学习与发现 - 少量高质量的数据来源优于大量多样性的数据来源 [25] - 对于 SFT 和知识蒸馏，基于答案过滤或验证答案似乎没有帮助 [26] - 较强的评估基准模型并不一定意味着它是一个更好的教师模型，例如，Quen 32B 是比 Deepseek R1 更好的教师模型 [21] - 通过知识蒸馏，模型可以在某些领域超越教师模型，例如在法律推理领域 [35][36][37] 实践建议 - 根据特定领域调整数据配方，从 Open Thoughts 的配方开始迭代 [29] - 针对代码、科学和数学等不同领域，应区别研究流水线的每个步骤 [29][30] - 如果特定领域的数据不足，可以将现有数据转换为问题，并使用上下文示例生成更多数据 [32] - 评估至关重要，需要使用 Evalchemy 等开源库来确保模型改进的有效性 [33][34]

Group 1 - The core viewpoint highlights the robust growth in the computer, communication, and electronic device manufacturing industry, with a year-on-year increase of 11.1% in added value for the first half of the year, surpassing the overall industrial growth rate by 4.7 percentage points [2] - The software and information technology service industry has shown solid progress in high-quality development, with business revenue reaching 557.88 billion yuan, a year-on-year growth of 11.2%, and total profits of 67.21 billion yuan, up 12.8% [2] - The hardware ecosystem is witnessing significant breakthroughs, with multiple domestic companies launching key products in high-growth areas such as security chips and AI-integrated devices, stimulating diverse consumer demand [2] Group 2 - The open-source ecosystem is evolving, with a focus on enhancing open-source software development, as evidenced by the introduction of several open-source projects and a growing community in fields like artificial intelligence and autonomous driving [3] - The open-source HarmonyOS ecosystem has expanded significantly, with over 1,200 product types and more than 1.19 billion devices equipped with the system, alongside a developer base exceeding 7.2 million [3] - The Ministry of Industry and Information Technology plans to promote digital transformation in the electronic information manufacturing industry and enhance the quality of new energy storage solutions [3][4]