Weekly Digest — 2026-W03

Fullstack app framework for web, desktop, and mobile.

小红书笔记 | 评论爬虫、抖音视频 | 评论爬虫、快手视频 | 评论爬虫、B 站视频 ｜ 评论爬虫、微博帖子 ｜ 评论爬虫、百度贴吧帖子 ｜ 百度贴吧评论回复爬虫 | 知乎问答文章｜评论爬虫

Autonomous AI development loop for Claude Code with intelligent exit detection

Collection of publicly available IPTV channels from all over the world

real time face swap and one-click video deepfake with only a single image

The Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra

Claude Code superpowers: core skills library

A command-line tool to download photos from iCloud

NVR with realtime local object detection for IP cameras

Source code for the X Recommendation Algorithm

📘 Home Assistant User documentation

Buzz transcribes and translates audio offline on your personal computer. Powered by OpenAI's Whisper.

The image geolocalization task aims to predict the location where an image was taken anywhere on Earth using visual clues. Existing large vision-language model (LVLM) approaches leverage world knowledge, chain-of-thought reasoning, and agentic capabilities, but overlook a common strategy used by humans -- using maps. In this work, we first equip the model Thinking with Map ability and formulate it as an agent-in-the-map loop. We develop a two-stage optimization scheme for it, including agentic reinforcement learning (RL) followed by parallel test-time scaling (TTS). The RL strengthens the agentic capability of model to improve sampling efficiency, and the parallel TTS enables the model to explore multiple candidate paths before making the final prediction, which is crucial for geolocalization. To evaluate our method on up-to-date and in-the-wild images, we further present MAPBench, a comprehensive geolocalization training and evaluation benchmark composed entirely of real-world images. Experimental results show that our method outperforms existing open- and closed-source models on most metrics, specifically improving Acc@500m from 8.0\% to 22.1\% compared to Gemini-3-Pro with Google Search/Map grounded mode.

Autoformalization, which translates natural language mathematics into formal statements to enable machine reasoning, faces fundamental challenges in the wild due to the multimodal nature of the physical world, where physics requires inferring hidden constraints (e.g., mass or energy) from visual elements. To address this, we propose MMFormalizer, which extends autoformalization beyond text by integrating adaptive grounding with entities from real-world mathematical and physical domains. MMFormalizer recursively constructs formal propositions from perceptually grounded primitives through recursive grounding and axiom composition, with adaptive recursive termination ensuring that every abstraction is supported by visual evidence and anchored in dimensional or axiomatic grounding. We evaluate MMFormalizer on a new benchmark, PhyX-AF, comprising 115 curated samples from MathVerse, PhyX, Synthetic Geometry, and Analytic Geometry, covering diverse multimodal autoformalization tasks. Results show that frontier models such as GPT-5 and Gemini-3-Pro achieve the highest compile and semantic accuracy, with GPT-5 excelling in physical reasoning, while geometry remains the most challenging domain. Overall, MMFormalizer provides a scalable framework for unified multimodal autoformalization, bridging perception and formal reasoning. To the best of our knowledge, this is the first multimodal autoformalization method capable of handling classical mechanics (derived from the Hamiltonian), as well as relativity, quantum mechanics, and thermodynamics. More details are available on our project page: MMFormalizer.github.io

A photorealistic and controllable 3D caricaturization framework for faces is introduced. We start with an intrinsic Gaussian curvature-based surface exaggeration technique, which, when coupled with texture, tends to produce over-smoothed renders. To address this, we resort to 3D Gaussian Splatting (3DGS), which has recently been shown to produce realistic free-viewpoint avatars. Given a multiview sequence, we extract a FLAME mesh, solve a curvature-weighted Poisson equation, and obtain its exaggerated form. However, directly deforming the Gaussians yields poor results, necessitating the synthesis of pseudo-ground-truth caricature images by warping each frame to its exaggerated 2D representation using local affine transformations. We then devise a training scheme that alternates real and synthesized supervision, enabling a single Gaussian collection to represent both natural and exaggerated avatars. This scheme improves fidelity, supports local edits, and allows continuous control over the intensity of the caricature. In order to achieve real-time deformations, an efficient interpolation between the original and exaggerated surfaces is introduced. We further analyze and show that it has a bounded deviation from closed-form solutions. In both quantitative and qualitative evaluations, our results outperform prior work, delivering photorealistic, geometry-controlled caricature avatars.

Large language models (LLMs) often fail to learn effective long chain-of-thought (Long CoT) reasoning from human or non-Long-CoT LLMs imitation. To understand this, we propose that effective and learnable Long CoT trajectories feature stable molecular-like structures in unified view, which are formed by three interaction types: Deep-Reasoning (covalent-like), Self-Reflection (hydrogen-bond-like), and Self-Exploration (van der Waals-like). Analysis of distilled trajectories reveals these structures emerge from Long CoT fine-tuning, not keyword imitation. We introduce Effective Semantic Isomers and show that only bonds promoting fast entropy convergence support stable Long CoT learning, while structural competition impairs training. Drawing on these findings, we present Mole-Syn, a distribution-transfer-graph method that guides synthesis of effective Long CoT structures, boosting performance and RL stability across benchmarks.

Reinforcement learning (RL) has emerged as a critical technique for enhancing LLM-based deep search agents. However, existing approaches primarily rely on binary outcome rewards, which fail to capture the comprehensiveness and factuality of agents' reasoning process, and often lead to undesirable behaviors such as shortcut exploitation and hallucinations. To address these limitations, we propose Citation-aware Rubric Rewards (CaRR), a fine-grained reward framework for deep search agents that emphasizes reasoning comprehensiveness, factual grounding, and evidence connectivity. CaRR decomposes complex questions into verifiable single-hop rubrics and requires agents to satisfy these rubrics by explicitly identifying hidden entities, supporting them with correct citations, and constructing complete evidence chains that link to the predicted answer. We further introduce Citation-aware Group Relative Policy Optimization (C-GRPO), which combines CaRR and outcome rewards for training robust deep search agents. Experiments show that C-GRPO consistently outperforms standard outcome-based RL baselines across multiple deep search benchmarks. Our analysis also validates that C-GRPO effectively discourages shortcut exploitation, promotes comprehensive, evidence-grounded reasoning, and exhibits strong generalization to open-ended deep research tasks. Our code and data are available at https://github.com/THUDM/CaRR.

Large language models (LLMs) are expected to be trained to act as agents in various real-world environments, but this process relies on rich and varied tool-interaction sandboxes. However, access to real systems is often restricted; LLM-simulated environments are prone to hallucinations and inconsistencies; and manually built sandboxes are hard to scale. In this paper, we propose EnvScaler, an automated framework for scalable tool-interaction environments via programmatic synthesis. EnvScaler comprises two components. First, SkelBuilder constructs diverse environment skeletons through topic mining, logic modeling, and quality evaluation. Then, ScenGenerator generates multiple task scenarios and rule-based trajectory validation functions for each environment. With EnvScaler, we synthesize 191 environments and about 7K scenarios, and apply them to Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) for Qwen3 series models. Results on three benchmarks show that EnvScaler significantly improves LLMs' ability to solve tasks in complex environments involving multi-turn, multi-tool interactions. We release our code and data at https://github.com/RUC-NLPIR/EnvScaler.

In real-world video question answering scenarios, videos often provide only localized visual cues, while verifiable answers are distributed across the open web; models therefore need to jointly perform cross-frame clue extraction, iterative retrieval, and multi-hop reasoning-based verification. To bridge this gap, we construct the first video deep research benchmark, VideoDR. VideoDR centers on video-conditioned open-domain video question answering, requiring cross-frame visual anchor extraction, interactive web retrieval, and multi-hop reasoning over joint video-web evidence; through rigorous human annotation and quality control, we obtain high-quality video deep research samples spanning six semantic domains. We evaluate multiple closed-source and open-source multimodal large language models under both the Workflow and Agentic paradigms, and the results show that Agentic is not consistently superior to Workflow: its gains depend on a model's ability to maintain the initial video anchors over long retrieval chains. Further analysis indicates that goal drift and long-horizon consistency are the core bottlenecks. In sum, VideoDR provides a systematic benchmark for studying video agents in open-web settings and reveals the key challenges for next-generation video deep research agents.

While humans develop core visual skills long before acquiring language, contemporary Multimodal LLMs (MLLMs) still rely heavily on linguistic priors to compensate for their fragile visual understanding. We uncovered a crucial fact: state-of-the-art MLLMs consistently fail on basic visual tasks that humans, even 3-year-olds, can solve effortlessly. To systematically investigate this gap, we introduce BabyVision, a benchmark designed to assess core visual abilities independent of linguistic knowledge for MLLMs. BabyVision spans a wide range of tasks, with 388 items divided into 22 subclasses across four key categories. Empirical results and human evaluation reveal that leading MLLMs perform significantly below human baselines. Gemini3-Pro-Preview scores 49.7, lagging behind 6-year-old humans and falling well behind the average adult score of 94.1. These results show despite excelling in knowledge-heavy evaluations, current MLLMs still lack fundamental visual primitives. Progress in BabyVision represents a step toward human-level visual perception and reasoning capabilities. We also explore solving visual reasoning with generation models by proposing BabyVision-Gen and automatic evaluation toolkit. Our code and benchmark data are released at https://github.com/UniPat-AI/BabyVision for reproduction.

We introduce Parallel Coordinated Reasoning (PaCoRe), a training-and-inference framework designed to overcome a central limitation of contemporary language models: their inability to scale test-time compute (TTC) far beyond sequential reasoning under a fixed context window. PaCoRe departs from the traditional sequential paradigm by driving TTC through massive parallel exploration coordinated via a message-passing architecture in multiple rounds. Each round launches many parallel reasoning trajectories, compacts their findings into context-bounded messages, and synthesizes these messages to guide the next round and ultimately produce the final answer. Trained end-to-end with large-scale, outcome-based reinforcement learning, the model masters the synthesis abilities required by PaCoRe and scales to multi-million-token effective TTC without exceeding context limits. The approach yields strong improvements across diverse domains, and notably pushes reasoning beyond frontier systems in mathematics: an 8B model reaches 94.5% on HMMT 2025, surpassing GPT-5's 93.2% by scaling effective TTC to roughly two million tokens. We open-source model checkpoints, training data, and the full inference pipeline to accelerate follow-up work.

Competitive programming presents great challenges for Code LLMs due to its intensive reasoning demands and high logical complexity. However, current Code LLMs still rely heavily on real-world data, which limits their scalability. In this paper, we explore a fully synthetic approach: training Code LLMs with entirely generated tasks, solutions, and test cases, to empower code reasoning models without relying on real-world data. To support this, we leverage feature-based synthesis to propose a novel data synthesis pipeline called SynthSmith. SynthSmith shows strong potential in producing diverse and challenging tasks, along with verified solutions and tests, supporting both supervised fine-tuning and reinforcement learning. Based on the proposed synthetic SFT and RL datasets, we introduce the X-Coder model series, which achieves a notable pass rate of 62.9 avg@8 on LiveCodeBench v5 and 55.8 on v6, outperforming DeepCoder-14B-Preview and AReal-boba2-14B despite having only 7B parameters. In-depth analysis reveals that scaling laws hold on our synthetic dataset, and we explore which dimensions are more effective to scale. We further provide insights into code-centric reinforcement learning and highlight the key factors that shape performance through detailed ablations and analysis. Our findings demonstrate that scaling high-quality synthetic data and adopting staged training can greatly advance code reasoning, while mitigating reliance on real-world coding data.

While the Transformer architecture dominates many fields, its quadratic self-attention complexity hinders its use in large-scale applications. Linear attention offers an efficient alternative, but its direct application often degrades performance, with existing fixes typically re-introducing computational overhead through extra modules (e.g., depthwise separable convolution) that defeat the original purpose. In this work, we identify a key failure mode in these methods: global context collapse, where the model loses representational diversity. To address this, we propose Multi-Head Linear Attention (MHLA), which preserves this diversity by computing attention within divided heads along the token dimension. We prove that MHLA maintains linear complexity while recovering much of the expressive power of softmax attention, and verify its effectiveness across multiple domains, achieving a 3.6\% improvement on ImageNet classification, a 6.3\% gain on NLP, a 12.6\% improvement on image generation, and a 41\% enhancement on video generation under the same time complexity.

Large Reasoning Models (LRMs) achieve remarkable performance by explicitly generating multi-step chains of thought, but this capability incurs substantial inference latency and computational cost. Collaborative inference offers a promising solution by selectively allocating work between lightweight and large models, yet a fundamental challenge remains: determining when a reasoning step requires the capacity of a large model or the efficiency of a small model. Existing routing strategies either rely on local token probabilities or post-hoc verification, introducing significant inference overhead. In this work, we propose a novel perspective on step-wise collaboration: the difficulty of a reasoning step can be inferred from its very first token. Inspired by the "Aha Moment" phenomenon in LRMs, we show that the entropy of the initial token serves as a strong predictor of step difficulty. Building on this insight, we introduce GlimpRouter, a training-free step-wise collaboration framework. GlimpRouter employs a lightweight model to generate only the first token of each reasoning step and routes the step to a larger model only when the initial token entropy exceeds a threshold. Experiments on multiple benchmarks demonstrate that our approach significantly reduces inference latency while preserving accuracy. For instance, GlimpRouter attains a substantial 10.7% improvement in accuracy while reducing inference latency by 25.9% compared to a standalone large model on AIME25. These results suggest a simple yet effective mechanism for reasoning: allocating computation based on a glimpse of thought rather than full-step evaluation.

微软想要让 Windows 的 AI 功能更醒目。根据 Windows 11 的预览版信息，它正在测试在资源管理器中集成 Copilot 的新功能，不是在右键菜单中添加一个“Ask Copilot”按钮，而是集成在资源管理器中，位于侧边栏或类似详细信息/预览窗格界面中。

根据 Netblocks 以及 Cloudflare Rader 的监测，伊朗全国断网四天。自 2025 年 12 月起，伊朗发生了一系列抗议活动，起因是民众对通货膨胀飙升、食品价格上涨以及伊朗里亚尔大幅贬值感到不满。示威活动最初由店主和市场商贩发起，进入新年后，抗议规模日益扩大。维基百科显示，目前有至少 2000 人死亡，逾万人被捕。

全球首台超临界二氧化碳发电机组“超碳一号”在贵州六盘水首钢水钢集团投入商业运行，这是全球范围内首次将超临界二氧化碳发电技术从实验室推向商业落地。不论是火电还是核电，原理都类似于“烧开水”，就是用热量将水变为水蒸气，推动汽轮机转动来发电。但超临界二氧化碳发电技术是一种革新型热电转换技术。这一技术是把温度超过 31 摄氏度、压力升高至 73 个大气压以上环境中的超临界二氧化碳作为循环工质，将其送进发电系统里，再通过压缩机和换热器提高超临界二氧化碳的压力和温度，让高温高压的二氧化碳推动透平旋转，进而产生电能。相比现役烧结余热蒸汽发电技术，“超碳一号”发电效率提升 85% 以上，净发电量提升 50% 以上。

Linus Torvalds 在 GitHub 上公开了名为 AudioNoise 的随机数字音效个人项目，他在 README 文件中透露使用了 Google 的 AI 驱动 IDE 工具 Antigravity 进行了辅助编程——或者叫 Vibe Coding。Antigravity 是 Windsurf 的分支，而 Windsurf 则是微软 VS Code 的分支。

因为大量用户在 APP 上与 AI 智能体“聊黄”，APP 的主要开发和运营者被追究了刑责。2025 年 9 月，上海市徐汇区人民法院一审判决，两名被告人犯制作淫秽物品牟利罪，分别获刑四年、一年半。此案成为国内首起 AI 服务提供者涉黄获刑的案件。案涉 APP Alien Chat（AC）是一款 AI 伴侣聊天应用，定位是为年轻群体提供亲密陪伴和情感支持。用户在 AC 注册会员后，可与 AI 聊天。判决书披露，AC App 手机注册用户 11.6 万人，其中付费用户 2.4 万人。截至案发，共收取会员充值费 363 万余元。两名被告人不服判决提出上诉，案件二审将于 1 月 14 日在上海市第一中级人民法院开庭。

在本该白雪皑皑的季节，喜马拉雅山却是光秃秃的，因为今年冬季的降雪量大幅减少。相比 1980-2020 年之间的平均降雪量，过去五年喜马拉雅山多数年份的降雪量都出现了下降。全球的气温上升也意味着少量积雪会很快融化。气象学家表示积雪减少不仅改变喜马拉雅山面貌，还会影响当地数亿人的生活和生态系统。融雪是该地区主要的淡水来源。印度气象部门记录显示 12 月印度北部几乎所有地区都没有降雨和降雪。根据数据集 ERA-5（European Centre for Medium-Range Weather Forecasts Reanalysis），相比 40 年长期平均水平（1980-2020 年），喜马拉雅山西北部的降雪量过去五年减少了 25%。

全国断网之后，伊朗居民主要通过 Starlink 设备与外界通信，将抗议视频传递出去。而 Starlink 也收到了严重干扰，时断时续。非营利组织 Miaan Group 的 Amir Rashidi 称伊朗政府开始搜查和没收 Starlink 设备。一位德黑兰用户通过 Starlink 接受了 WSJ 的采访，表示自己上传了亲戚拍摄的抗议视频，发送给了国外第三方，由他们发布到社媒上。Starlink 连接状况通常早上或中午时好点。Starlink 终端在伊朗属于非法设备，是通过走私进来的。在 2022 年上一次大规模抗议之后，Starlink 终端大量涌入伊朗。NetFreedom Pioneers 等组织向伊朗运送了数千套 Starlink 设备。

应用脑机接口技术已突破英语语音和文字合成，但针对汉语解码的脑机接口技术研究相对较少。中国科学院研究人员针对汉语解码，开发出植入式高通量柔性脑机接口系统和汉语言实时神经网络解码算法，首次实现脑机接口实时汉语解码和语句合成。相比于英语，汉语具有其独特性。具体而言，英语是以多音节为主的非声调语言，汉语则是以单音节为主的声调语言。同时，英语词汇量较大，常用英语单词约为 20000 个，而汉语通过约 400 个汉语音节加 4 个声调，可构建出覆盖日常需求的 3500 多个常用汉字。研究团队利用汉语本身优势，从约 400 个汉语音节和 4 个声调入手，将其作为稳定的中间解码单元，实现从脑电到文字的“翻译”，且通过解码这些汉语音节和声调，可外推至全部汉字。研究显示，受试者经过9天的语言解码任务后，394个汉语音节（解码未覆盖音节为生僻音节且受试者不认识）纯神经解码平均准确率达到71.2%，单音节解码延时65ms，实时汉语语句解码速率达到 49.6 字/分钟。

苹果对下架应用的措施很模糊，留下了可伸缩的空间。相比下 Google 则制定了明确的规定，根据其支持文档：“我们禁止任何应用包含或宣传色情内容或粗言秽语，包括淫秽内容或以性取悦为目的的任何内容或服务。我们禁止任何疑似宣传或招揽有偿性行为的应用或应用内容。我们禁止任何应用包含或宣传与性掠夺行为相关的内容，或在未经当事人同意的情况下散布色情内容。”xAI 的 Grok AI 最近的行为完全符合该规定，Web 版本的 Grok 已经将 deepfake 功能限制为付费用户使用，但 Grok 应用仍然没有任何限制。对于为何没有下架 Grok 应用 Google 至今拒绝置评。

挪威公布了 2025 年 12 月和全年汽车销量数据，纯电和插电混动汽车销量占到了总销量的 97.5%，基本实现了它在 2017 年设定的到 2025 年停售燃油汽车的目标。2025 年挪威共注册新乘用车 179,549 辆。其中纯电 172,232 辆，插电混动 2,751 辆，传统燃油汽车 2,306 辆，柴油车 1,773 辆，汽油车 487 辆。燃油车主要包括特殊用途车辆如无障碍汽车或警用及其它应急车辆。中国电动汽车在挪威的销量也实现了增长，中国品牌的份额从去年的 10.4% 上升至 13.7%。

苹果与 Google 达成多年协议，将使用 Gemini 驱动其智能助手 Siri。苹果早在 2024 年就承诺推出 AI 驱动的 Siri 助手，但因为内部大模型不可靠而一直延期。苹果之后决定选择外部 AI 模型，它测试了 OpenAI 的 ChatGPT 和 Anthropic 的 Claude，最终决定选择了 Google 的 Gemini。报道称，苹果将为此每年向 Google 支付 10 亿美元。考虑到 Google 为了获得苹果设备的默认搜索引擎位置而每年支付超过 200 亿美元（2022 年的数字，如今可能会更高），10 亿美元对苹果而言只是少收点钱而已。新版本的 Siri 将在今年晚些时候在 iOS 26、iPadOS 26 和 macOS 26 Tahoe 上推出。

中国公司向国际电信联盟（ITU）申请发射逾 20 万颗互联网卫星。去年 12 月 30 日成立的无线电频谱开发利用和技术创新研究院申请了规模最大的两个项目——CTC-1 和 CTC-2——分别申请发射 96,714 颗卫星，它在成立的第二天递交了申请。除此之外，中国移动的 L1 项目包含了 2520 颗卫星，千帆星座的开发商上海垣信卫星科技有限公司递交了 1296 颗卫星计划。千帆星座计划到 2030 年部署 1.5 万颗卫星，国网计划发射 1.3 万颗卫星。向国际电信联盟递交申请的还有：国电高科天启 3G（1,132 颗卫星）、中国移动 M1（144 颗卫星）、航天驭星 YX-5（106 颗卫星）、GalaxySpace 的 Galaxy-SAR-2 网络（96 颗卫星）和 BlackSpider-3 星座（81 颗卫星）。