OrangeBot.AI Digest — 2025-12-31

Get Spotify tracks in true FLAC from Tidal, Qobuz & Amazon Music — no account required.

Introduction to Machine Learning Systems

Get 10X more out of Claude Code, Codex or any coding agent

MCP server and Claude plugin for Postgres skills and documentation. Helps AI coding tools generate better PostgreSQL code.

🎯 告别信息过载，AI 助你看懂新闻资讯热点，支持 RSS 订阅，简单的舆情监控分析 - 多平台热点聚合+基于 MCP 的AI分析工具。监控35个平台（抖音、知乎、B站、华尔街见闻、财联社等），智能筛选+自动推送+AI对话分析（用自然语言深度挖掘新闻：趋势追踪、情感分析、相似检索等20种工具）。支持企业微信/个人微信/飞书/钉钉/Telegram/邮件/ntfy/bark/slack 推送，30秒快速部署，1分钟手机通知，无需编程。支持Docker部署，支持数据远程云存储⭐ 让算法为你服务，用AI理解热点

Examples and guides for using the OpenAI API

🍃 Organic Maps is a free Android & iOS offline maps app for travelers, tourists, hikers, and cyclists. It uses crowd-sourced OpenStreetMap data and is developed with love by the community. No ads, no tracking, no data collection, no crapware. Please donate to support the development!

SoTA open-source TTS

🔥 🔥 🔥 Open Source Airtable Alternative

In this report, we introduce UltraShape 1.0, a scalable 3D diffusion framework for high-fidelity 3D geometry generation. The proposed approach adopts a two-stage generation pipeline: a coarse global structure is first synthesized and then refined to produce detailed, high-quality geometry. To support reliable 3D generation, we develop a comprehensive data processing pipeline that includes a novel watertight processing method and high-quality data filtering. This pipeline improves the geometric quality of publicly available 3D datasets by removing low-quality samples, filling holes, and thickening thin structures, while preserving fine-grained geometric details. To enable fine-grained geometry refinement, we decouple spatial localization from geometric detail synthesis in the diffusion process. We achieve this by performing voxel-based refinement at fixed spatial locations, where voxel queries derived from coarse geometry provide explicit positional anchors encoded via RoPE, allowing the diffusion model to focus on synthesizing local geometric details within a reduced, structured solution space. Our model is trained exclusively on publicly available 3D datasets, achieving strong geometric quality despite limited training resources. Extensive evaluations demonstrate that UltraShape 1.0 performs competitively with existing open-source methods in both data processing quality and geometry generation. All code and trained models will be released to support future research.

Recently unified generation and editing models have achieved remarkable success with their impressive performance. These models rely mainly on text prompts for instruction-based editing and generation, but language often fails to capture users intended edit locations and fine-grained visual details. To this end, we propose two tasks: scribble-based editing and generation, that enables more flexible creation on graphical user interface (GUI) combining user textual, images, and freehand sketches. We introduce DreamOmni3, tackling two challenges: data creation and framework design. Our data synthesis pipeline includes two parts: scribble-based editing and generation. For scribble-based editing, we define four tasks: scribble and instruction-based editing, scribble and multimodal instruction-based editing, image fusion, and doodle editing. Based on DreamOmni2 dataset, we extract editable regions and overlay hand-drawn boxes, circles, doodles or cropped image to construct training data. For scribble-based generation, we define three tasks: scribble and instruction-based generation, scribble and multimodal instruction-based generation, and doodle generation, following similar data creation pipelines. For the framework, instead of using binary masks, which struggle with complex edits involving multiple scribbles, images, and instructions, we propose a joint input scheme that feeds both the original and scribbled source images into the model, using different colors to distinguish regions and simplify processing. By applying the same index and position encodings to both images, the model can precisely localize scribbled regions while maintaining accurate editing. Finally, we establish comprehensive benchmarks for these tasks to promote further research. Experimental results demonstrate that DreamOmni3 achieves outstanding performance, and models and code will be publicly released.

We formulate long-context language modeling as a problem in continual learning rather than architecture design. Under this formulation, we only use a standard architecture -- a Transformer with sliding-window attention. However, our model continues learning at test time via next-token prediction on the given context, compressing the context it reads into its weights. In addition, we improve the model's initialization for learning at test time via meta-learning at training time. Overall, our method, a form of Test-Time Training (TTT), is End-to-End (E2E) both at test time (via next-token prediction) and training time (via meta-learning), in contrast to previous forms. We conduct extensive experiments with a focus on scaling properties. In particular, for 3B models trained with 164B tokens, our method (TTT-E2E) scales with context length in the same way as Transformer with full attention, while others, such as Mamba 2 and Gated DeltaNet, do not. However, similar to RNNs, TTT-E2E has constant inference latency regardless of context length, making it 2.7 times faster than full attention for 128K context. Our code is publicly available.

We systematically evaluate Parameter-Efficient Fine-Tuning (PEFT) methods under the paradigm of Reinforcement Learning with Verifiable Rewards (RLVR). RLVR incentivizes language models to enhance their reasoning capabilities through verifiable feedback; however, while methods like LoRA are commonly used, the optimal PEFT architecture for RLVR remains unidentified. In this work, we conduct the first comprehensive evaluation of over 12 PEFT methodologies across the DeepSeek-R1-Distill families on mathematical reasoning benchmarks. Our empirical results challenge the default adoption of standard LoRA with three main findings. First, we demonstrate that structural variants, such as DoRA, AdaLoRA, and MiSS, consistently outperform LoRA. Second, we uncover a spectral collapse phenomenon in SVD-informed initialization strategies (e.g., PiSSA, MiLoRA), attributing their failure to a fundamental misalignment between principal-component updates and RL optimization. Furthermore, our ablations reveal that extreme parameter reduction (e.g., VeRA, Rank-1) severely bottlenecks reasoning capacity. We further conduct ablation studies and scaling experiments to validate our findings. This work provides a definitive guide for advocating for more exploration for parameter-efficient RL methods.

The issue localization task aims to identify the locations in a software repository that requires modification given a natural language issue description. This task is fundamental yet challenging in automated software engineering due to the semantic gap between issue description and source code implementation. This gap manifests as two mismatches:(1) symptom-to-cause mismatches, where descriptions do not explicitly reveal underlying root causes; (2) one-to-many mismatches, where a single issue corresponds to multiple interdependent code entities. To address these two mismatches, we propose GraphLocator, an approach that mitigates symptom-to-cause mismatches through causal structure discovering and resolves one-to-many mismatches via dynamic issue disentangling. The key artifact is the causal issue graph (CIG), in which vertices represent discovered sub-issues along with their associated code entities, and edges encode the causal dependencies between them. The workflow of GraphLocator consists of two phases: symptom vertices locating and dynamic CIG discovering; it first identifies symptom locations on the repository graph, then dynamically expands the CIG by iteratively reasoning over neighboring vertices. Experiments on three real-world datasets demonstrates the effectiveness of GraphLocator: (1) Compared with baselines, GraphLocator achieves more accurate localization with average improvements of +19.49% in function-level recall and +11.89% in precision. (2) GraphLocator outperforms baselines on both symptom-to-cause and one-to-many mismatch scenarios, achieving recall improvement of +16.44% and +19.18%, precision improvement of +7.78% and +13.23%, respectively. (3) The CIG generated by GraphLocator yields the highest relative improvement, resulting in a 28.74% increase in performance on downstream resolving task.

Mixture-of-Experts (MoE) architectures have advanced the scaling of Large Language Models (LLMs) by activating only a sparse subset of parameters per input, enabling state-of-the-art performance with reduced computational cost. As these models are increasingly deployed in critical domains, understanding and strengthening their alignment mechanisms is essential to prevent harmful outputs. However, existing LLM safety research has focused almost exclusively on dense architectures, leaving the unique safety properties of MoEs largely unexamined. The modular, sparsely-activated design of MoEs suggests that safety mechanisms may operate differently than in dense models, raising questions about their robustness. In this paper, we present GateBreaker, the first training-free, lightweight, and architecture-agnostic attack framework that compromises the safety alignment of modern MoE LLMs at inference time. GateBreaker operates in three stages: (i) gate-level profiling, which identifies safety experts disproportionately routed on harmful inputs, (ii) expert-level localization, which localizes the safety structure within safety experts, and (iii) targeted safety removal, which disables the identified safety structure to compromise the safety alignment. Our study shows that MoE safety concentrates within a small subset of neurons coordinated by sparse routing. Selective disabling of these neurons, approximately 3% of neurons in the targeted expert layers, significantly increases the averaged attack success rate (ASR) from 7.4% to 64.9% against the eight latest aligned MoE LLMs with limited utility degradation. These safety neurons transfer across models within the same family, raising ASR from 17.9% to 67.7% with one-shot transfer attack. Furthermore, GateBreaker generalizes to five MoE vision language models (VLMs) with 60.9% ASR on unsafe image inputs.

随着特朗普政府日益强化入境审查，甚至要倒查五年的社交媒体历史记录，外国科技从业者、研究人员和会议主讲越来越多的避开美国。行业会议和活动逐渐转移到更友好的欧洲、加拿大和亚洲地区。专栏作家 Steven Vaughan-Nichols 今年参加了 13 场科技会议，大部分是在美国之外举行的。在这些会议上，非美国人最关心的非科技话题都围绕着特朗普第二个任期给美国带来的巨大变化，与会者多表示他们不会去美国找工作，也不会去美国参加会议。展会主办方也开始取消原计划在美国举办的展会，转移到欧洲、加拿大和亚洲。曾经精英们为去美国愿意放弃一切，但如今情势出现了逆转。

以色列部署了无人机激光防御系统 Iron Beam。Iron Beam 功率达到 10 万瓦，能以极低的单次发射成本击落无人机、火箭弹和迫击炮弹。它由以色列拉斐尔先进防御系统（Rafael Advanced Defense Systems）研发，其核心是先进激光源和独一无二的光电瞄准系统，能在更远的作战距离上以高精度和高效率拦截各种目标。以色列没有透露太多细节，只是强调 Iron Beam 已成功拦截火箭弹、迫击炮弹和无人机，其投入使用标志着高能激光防御时代的开始。

根据印度政府的年终经济评估，该国 GDP 超过日本成为全球第四大经济体。按照目前的增长速度，印度 GDP 预计会在三年内超过德国成为全球第三大经济体，仅次于美国和中国。印度 GDP 今年估计为 4.18 万亿美元，预计 2030 年达到 7.3 万亿美元。印度 2025 年 GDP 正式数据要到 2026 年公布。国际货币基金组织的预测是印度 GDP 将在明年超过日本。

冰岛经历了有记录以来最暖和的平安夜，最高气温达到了 19.8C。冰岛气象局称，东部小镇 Seyðisfjörður 气温高达 19.8C，东部 Borgarfjörður 的 Bakkagerði 气温达到了 19.7C。12 月冰岛平均气温在 -1C 到 4C 之间。气象学家表示原因是热带暖气流笼罩了冰岛。今年以来冰岛各地遭遇了破纪录的热浪，首次在野外发现蚊子。此前冰岛是南极洲之外唯一一个没有野蚊子的地方。

2024 年中国企业开展的癌症治疗药物临床试验数量连续两年超过美国，居世界首位。统计数据显示，总部设在中国的企业开展的癌症临床试验数量在 2024 年为 896 项，约占全球整体的 39%。超过约为 32% 的美国，居世界首位。远超过欧洲（约 20%）和日本（约 4%）。 2023 年中国（约35%）以微弱优势首次超过美国（约34%），2024 年优势进一步拉大。而 2009 年中国企业在癌症领域的临床试验数量仅占全球的 2% 左右。百济神州董事长兼首席执行官欧雷强（John V. Oyler）指出，“全球 25% 的新增癌症患者在中国就诊。在推进癌症领域的业务和研究方面，中国的存在是不可缺少的”。

自由软件 Android 应用商店 F-Droid 宣布升级了其核心服务器硬件，显著加快了构建和发布更新的速度。F-Droid 表示它的服务器不是托管在某个不知位置不知道员工身份的数据中心，而是交给一位长期贡献者管理，他们知道位置，能远程控制，也知道维护者身份。此前使用的服务器硬件有 12 年历史运行了 5 年，升级到新服务器之后，发布更新的速度从 1-9 月每 3-4 天发布一次更新加快到 10 月每 2 天更新一次，11 月每天一次，12 月每天更新两次。

根据发表在《Archives of Sexual Behavior》期刊上的一项研究，俄罗斯入侵乌克兰改变了乌克兰人的成人内容消费习惯。调查显示，2022 年 3 月初乌克兰人的网络行为发生显著改变。这一时间段与俄罗斯入侵的时间点高度吻合，俄罗斯入侵始于 2 月 24 日。数据显示，期间乌克兰民众对色情内容的搜索量显著增加，对战争地图和保持社交距离信息的搜索量也大幅上升。战争严重程度与成人网站访问量之间存在显著的统计相关性。日益加剧的孤立感和危险感似乎推动了对色情内容兴趣的增长。研究人员称，在集体威胁和社会动荡的紧张时期，人们可能会转向独处的性行为如观看色情作为一种应对或自我调节的策略。

MongoDB 服务器软件在圣诞节前曝出了一个高危漏洞，影响自 2017 年以来发布的所有版本。该漏洞被称为 MongoBleed aka CVE-2025-14847。MongoDB 在 12 月 24 日释出了补丁，声称没有证据表明有人利用该漏洞。12 月 27 日，育碧热门游戏《彩虹六号：围攻X》的服务器遭到入侵，黑客向全服赠送了 21 亿游戏虚拟货币以及极其罕见的皮肤，其中包括开发者专属皮肤。此次攻击迫使育碧将整个游戏服务器下线，直到 12 月 29 日才重新上线，回滚黑客送出的虚拟币和皮肤。有报道称，黑客就是利用刚刚曝出但育碧没有及时修复的 MongoDB 漏洞入侵和接管游戏系统，同时还窃取了育碧几乎所有游戏的源代码——但这一消息尚未得到育碧官方证实。

美国海洋勘探公司 Ocean Infinity 准备对失踪 12 年的 MH370 展开新一轮搜寻。从吉隆坡飞往北京的 MH370 航班于 2014 年 3 月 8 日失踪，当时机上有 239 人。Ocean Infinity 与马来西亚政府签署了一份“无发现不收费”的协议，Ocean Infinity 只有在找到飞机残骸后才能获得 1.102 亿美元的报酬。它在 2018 年曾达成类似协议，但三个月搜寻一无所获。最新的搜寻行动预计将持续 55 天，针对的是南印度洋的一个偏远区域，将使用改进的声纳和分析技术。

Meta 以大约 20 亿美元的价格收购了中资背景的 AI 智能体初创公司 Manus。这个价格是 Manus 下一轮融资寻求的估值。Manus 母公司是总部位于新加坡的蝴蝶效应（Butterfly Effect），它在今年 3 月发布了一鸣惊人的智能体演示视频而引发广泛关注。它的投资者包括了 Benchmark、腾讯、真格基金和红杉资本等。Manus 对其 AI 模型的访问收取了 39 美元或 199 美元的高价，它最近声称有数百万用户，年度经常性收入突破了 1 亿美元。Meta 就是在此时与 Manus 展开了收购谈判。

KDE 开发者总结了桌面环境 Plasma 在 2025 年的重要进展：切换到 Wayland 显示服务器的工作基本完成，2027 年初发布的 Plasma 将停止支持 X11 会话；Plasma 持续改进和成熟，成为众多面向游戏发行版的默认桌面环境，这些发行版包括了 Bazzite、CachyOS、Garuda、Nobara，以及 Valve 掌机/主机运行的 SteamOS。Fedora 发行版也将其 Plasma 桌面版本与 GNOME 桌面版本放在同等位置，唯一能在苹果新 Mac 设备上运行的发行版 Asahi Linux 使用的也是 KDE Plasma 桌面。Parrot Linux 最近也开始默认使用 Plasma。EndeavourOS、Manjaro、NixOS、OpenMandriva、Slackware 和 TuxedoOS 等老牌发行版的默认桌面环境都是 Plasma。

加拿大麦吉尔大学与美国德雷塞尔大学团队联合开发出一种颇具创意的高分辨率3D打印新技术。他们将雌性蚊子的口器（吸血管）转化成了高分辨率的3D打印喷嘴。这种技术不仅能打印出精度达 20 微米的极细线条，还为解决昂贵、高能耗的微纳制造难题提供了可持续的生物学方案。高分辨率 3D 打印对喷嘴精度要求极高。目前市售的超细喷嘴多由特种金属或玻璃制成，制造工艺复杂，成本高昂。研究团队指出，传统喷嘴在生产和使用过程中不仅产生大量环境废弃物，还可能因工艺局限带来健康风险。为了寻找替代方案，研究团队将目光投向自然界中高度进化的微结构——蚊子口器。经过数百万年进化，蚊子口器形成了一种直径仅为人类发丝直径一半左右的天然微针结构，兼具特殊几何形态和力学韧性。研究团队在显微镜下分离出蚊子吸血管，并利用特种树脂将其固定在标准塑料分配器尖端。结果发现，这种生物喷嘴能承受极大的压力，打印出的复杂结构精细程度大约是目前商业打印喷嘴的 2 倍。

网信办发布了《人工智能拟人化互动服务管理暂行办法（征求意见稿）》，意见截止日期 1 月 25 日。该《暂行办法》包含了被认为全球最严厉的政策，要求服务商采取措施阻止 AI 帮助用户自杀或自残。《暂行办法》包括： 第八条 提供者应当落实拟人化互动服务安全主体责任，建立健全算法机制机理审核、科技伦理审查、信息发布审核、网络安全、数据安全、个人信息保护、反电信网络诈骗、重大风险预案、应急处置等管理制度，具有安全可控的技术保障措施，配备与产品规模、业务方向和用户群体相适应的内容管理技术和人员。 第九条 提供者应当在拟人化互动服务全生命周期履行安全责任，明确设计、运行、升级、终止服务等各阶段安全要求，保证安全措施与服务功能同步设计、同步使用，提升内生安全水平，加强运行阶段安全监测和风险评估，及时发现纠正系统偏差、处置安全问题，依法留存网络日志。提供者应当具备心理健康保护、情感边界引导、依赖风险预警等安全能力，不得将替代社会交往、控制用户心理、诱导沉迷依赖等作为设计目标。 第十一条 提供者应当具备用户状态识别能力，在保护用户个人隐私前提下，评估用户情绪及对产品和服务的依赖程度，发现用户存在极端情绪和沉迷的，采取必要措施予以干预。提供者应当预设回复模板，发现涉及威胁用户生命健康和财产安全的高风险倾向的，及时输出安抚和鼓励寻求帮助等内容，并提供专业援助方式。提供者应当建立应急响应机制，发现用户明确提出实施自杀、自残等极端情境时，由人工接管对话，并及时采取措施联络用户监护人、紧急联系人。针对未成年人、老年人用户，提供者应当在注册环节要求填写用户监护人、紧急联系人等信息。 第十七条 用户连续使用拟人化互动服务超过2个小时的，提供者应当以弹窗等方式动态提醒用户暂停使用服务。

中国车企的全球销量在 2025 年超过日本，首次跃居首位。根据2025 年 1～11 月各企业发布的资料和标普全球汽车（S&P Global Mobility）的数据，中国汽车的全球销量预计同比增长 17%，增至约 2700 万辆。中国在 2023 年首次位居汽车出口首位。整体销量也将在 2025 年跃居首位。日本车企合计销量约为 2500 万辆，与上年持平。过去世界汽车销售由美国和日本展开竞争。在顶峰时期的 2018 年日本销量近 3000 万辆。另一方面，中国国内的供应过剩迹象增强，最大车企比亚迪开始降价，价格竞争日趋激烈。中国汽车制造商正在转向出口寻找出路。

对尼尔森最新数据的分析显示，2025 年没有一部新的原创剧能进入十大最受欢迎的流媒体节目之列。这是尼尔森自 2020 年以来发布流媒体数据以来首次出现该情况。数据还显示，由广告支持的免费流媒体服务增长速度超过了付费流媒体服务。YouTube 是美国电视上观看量最高的流媒体服务，超过了 Netflix 和亚马逊总和。Netflix 在热门剧上仍然具有优势，在尼尔森每周十大热门原创节目榜单中占了约三分之二。但其主导地位正逐渐消失——该公司的流媒体观看份额占比降至 20% 以下。迪士尼流媒体服务份额三年以来停滞不前，而亚马逊则在迎头赶上。2025 年观看量最高的原创剧是《鱿鱼游戏》终季，之后是《星期三》第二季和《爱情岛》的最新季。