Python · ★ 57,010 · 🍴 18,145 · 📈 3,860 stars today
The job search that runs on your machine. AI job application framework built on Claude Code: evaluate postings, tailor CVs, write cover letters, prep interviews. Fork it and own it.
中文介绍 AI工作搜索平台，基于Claude Code，可评估职位、定制简历、撰写求职信、准备面试。适合个人或团队进行职位搜索与申请。
TypeScript · ★ 217,728 · 🍴 29,104 · 📈 1,310 stars today
The open source coding agent.
中文介绍 开源的代码代理工具，适用于自动化代码生成和优化。
Rust · ★ 11,274 · 🍴 799 · 📈 666 stars today
Solution for long term memory for agent coding CLIs and to facilitate handoff between different agent vendors
中文介绍 用于智能代理CLI的长期记忆解决方案，促进不同代理供应商之间的交接。
Python · ★ 139,649 · 🍴 16,579 · 📈 1,011 stars today
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
中文介绍 强大的模块化扩散模型GUI、API和后端，拥有图/节点界面，提供本地推理引擎。适合进行快速模型推理。
TypeScript · ★ 77,139 · 🍴 9,101 · 📈 564 stars today
🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
中文介绍 智能多玩家群体部署工具，协调自主工作流程，构建对话式AI系统，具有自适应记忆和自我学习能力。
TypeScript · ★ 98,172 · 🍴 6,260 · 📈 1,591 stars today
Orca is the place for working with a fleet of agents (e.g. Grok Bot, Claude Code). Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.
中文介绍 提供多个代理（如Grok Bot，Claude Code）的工作环境，可在桌面、移动和远程运行时使用，适合运行编码代理。
Python · ★ 76,200 · 🍴 5,922 · 📈 317 stars today
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
中文介绍 压缩工具输出、日志、文件和RAG块，减少编码代理的令牌使用，适合需要减少LLM令牌消耗的场景。
JavaScript · ★ 6,083 · 🍴 998 · 📈 301 stars today
中文介绍 操作系统分类系统，具体用途未知。
Python · ★ 2,864 · 🍴 212 · 📈 502 stars today
Video restoration tool
中文介绍 视频修复工具，适用于视频质量的提升。
Python · ★ 55,187 · 🍴 4,977 · 📈 607 stars today
"CLI-Anything: Making ALL Software Agent-Native" -- CLI-Hub: https://clianything.cc/
中文介绍 将所有软件转变为代理原生，通过CLI-Hub提供CLI工具。
TypeScript · ★ 152,613 · 🍴 27,136 · 📈 167 stars today
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
中文介绍 终端中的代理编码工具，理解代码库，通过执行常规任务、解释复杂代码和处理git工作流来提高编码速度。
Python · ★ 41,603 · 🍴 6,919 · 📈 355 stars today
Kronos: A Foundation Model for the Language of Financial Markets
中文介绍 金融市场语言的基础模型，用于理解金融市场的语言和趋势。
👍 26
Multimodal large language models (MLLMs) answer questions about whole images well, yet they still struggle with fine-grained region understanding. A key reason is that text-only supervision never specifies which regions support an answer. To close this gap, we present Xiaomi-Lens-0, a 4B MLLM built
👍 5
Multi-turn image editing aims to modify an image according to a sequence of instructions, making it better aligned with human-AI collaborative creation than single-turn editing. However, constructing high-quality multi-turn training data remains challenging, as conventional sequential rollout accumu
👍 49
Generating complete, coherent, and usable 3D indoor scenes from natural language requires strong design intelligence: the ability to design high-quality scenes and critique their quality. However, existing approaches either rely on general-purpose models lacking domain-specific design experience or
👍 23
Making a conference poster is visual coding in miniature: an entire paper must be compressed onto a single page, which exercises summarization, visual design, and code implementation at once, using nothing beyond the paper itself. Yet benchmarks for this task in the context of visual coding remain s
👍 17
AI agents increasingly use files and software tools to produce professional deliverables. Such work draws on materials, decisions, and feedback accumulated across assignments. We introduce OfficeTown, a generative simulation framework for professional organizations in which tasks and their context a
👍 9
Looped language models reuse physical Transformer blocks across logical depth, but established architectures entangle recurrence placement, input injection, and recurrent state organization. We isolate these axes in models trained from scratch at matched parameter scale, logical depth, and token bud
👍 16
Recent closed-source image generation systems can produce diverse visual content from simple compositions to information-dense graphics, yet open-weight text-to-image models still struggle with dense text and intricate layouts. To narrow this gap, we introduce MOSS-Image, an open generation system t
👍 22
A memory test for a video world model should say whether the scene was set, whether the asked change was carried out, and whether the last frame shows the result. Current tests fold these into one number, so a low score is easy to read as forgetting. When the change is not carried out, the hidden st
👍 2
Enhancing the agentic capabilities of Large Language Models (LLMs) to tackle complex software engineering (SWE) and general terminal-usage problems is a pressing challenge. Current state-of-the-art LLMs often go through large-scale agentic Reinforcement Learning (RL). However, they mostly rely on hi
👍 14
Creative intent in video editing can be expressed through text instructions (e.g., "remove the people"), image references (e.g., subject appearance), video references (e.g., camera movement), or a combination of the above. We introduce OmniXRef, a unified framework for multimodal multi-reference vid
👍 2
Reasoning over long document contexts remains a challenge for large language models (LLMs). Standard retrieval methods are inherently reactive, bottlenecking agents with complex, cross-document reasoning at inference time, while precomputed structures like knowledge graphs often fall short because t
👍 4
Text-based world models have recently shown potential to simulate complex RL environment state transitions for a diverse set of use cases. However, no standardized benchmark exists to compare, study and improve these models. Through this paper, we attempt to solve this problem by building GlobeBench
👍 2
AI agents increasingly browse the web on users' behalf, yet they consume webpages as screenshots or text, exposing their LLMs to unvetted third-party content. Browsers enforce semantics under which what an element can do depends on what it is and where it comes from, not on what it says. Agents flat
👍 32
The vocabulary of a language model is its native semantics: the embedding and the unembedding fix the meaning of a state at the two ends of the residual stream. Between them, existing methods either reuse a fixed lens or treat the states as a point cloud, and neither reads a state the way the model
👍 144
We study how vision-language models (VLMs) can learn to think and actively explore in 3D space. Instead of providing multi-view images, videos, or 3D representations to the model, we preserve the native visual encoding of VLMs, i.e., a single image, one at a time, just like humans, and treat the 3D
👍 11
Large language model (LLM) agents can keep scientific experiments running for months with little human supervision, but reliable autonomous research also requires scientific judgment. Evaluating scientific judgment is difficult because research decisions are open-ended and depend on project context.
👍 14
Retrieval is a fundamental task in multimodal systems: given an image or text query, the goal is to find the most relevant images or texts in a collection, often using a shared embedding space. Reliable retrieval requires these embeddings to capture not only which concepts are present, but also how
👍 8
Peer review is under strain from the growth of submissions, and review comments written by humans or AI can misstate what a paper says. We introduce ReviewFact, a benchmark that labels such factual errors in review comments under one definition across NLP, machine learning, and the natural sciences,
👍 3
A coding agent works on repositories that can be far larger than its context window can hold. Prompt compressors and retrievers offer solutions to this problem, but patching an issue needs two things at once: the files that must change, and their code exactly as written, since the model must reprodu
👍 3
Self-evolving language models train on tasks they generate, so every round must decide which updated model the next round inherits. Current loops such as R-Zero, Agent0, and Skill Self-Play make this decision through the task, admitting tasks that look learnable and inheriting every resulting update
👍 2
Transcribing expressive piano performances into scores requires interpreting individual note timings within an underlying metrical structure. However, score attributes are associated with notes, whereas beats may occur without coincident note onsets. We introduce DuET, a Dual-Encoder Transformer tha
👍 7
Speculative decoding accelerates autoregressive generation by drafting multiple tokens per step, yet feature-based drafters rely on target hidden features whose contextual emphasis may not align with the needs of subsequent draft positions. Hybrid linear attention (HLA) models maintain compact recur
👍 4
Large language models (LLMs) hold broad knowledge in one set of weights, but lack interpretability and trustworthy reasoning, especially in high-stakes use cases. A neuro-symbolic model, with its interpretable architecture and System 2 reasoning capacity, is a promising alternative if built at the s
👍 4
Long-horizon agents often struggle to recover from failures whose causes lie several steps earlier, beyond what local reflection or backtracking can resolve. We introduce BELIEFWEAVER, a framework that combines persistent evidence across attempts with prospective reasoning over repair directions. It
👍 7
Web application development often proceeds through successive requests that extend an existing implementation. Yet web-generation benchmarks primarily evaluate development toward fixed specifications. We introduce EvoWeb, a novel benchmark comprising 320 cumulative requests over 32 web applications,
👍 6
Reliable deployment of LLM-based multi-agent systems calls for runtime auditors that predict terminal outcomes from observed execution prefixes. These auditors typically learn from a limited corpus of trajectories with verified outcomes. Expanding this corpus requires costly execution and verificati
👍 110
Evaluating agents by outcomes alone can obscure the capabilities that produce them. This problem is especially pronounced in evolving environments, where outcomes reflect a closed-loop interaction between agent behavior and changing external conditions. We introduce LiveMACEBench, a process-aware be
👍 5
Repository-level software engineering (SWE) is a challenging long-horizon setting: agents must reason over extended interactions, use tools, and adapt to stateful environments. Recent work trains SWE agents with reinforcement learning methods such as Group Relative Policy Optimization (GRPO), which
👍 15
Large language models (LLMs) progressively revise their predictions across Transformer layers, yet we typically observe only the final output, discarding the trajectory through which it is formed. We introduce Belief-Trajectory Energy(BTE), a model-grounded measure that characterizes an input throug
👍 10
On-policy distillation (OPD) improves language-model by supervising a student on prefixes sampled from its own policy, but its supervision remains confined to the output space. Consequently, OPD transfers what a teacher predicts without supervising the intermediate computation that produces those pr
@rahulchhabra07 · 9.8K 粉丝 · 1.2M 阅 · 1.5K 赞 · 360 转
In 1924 Hans Berger invented EEG to build telepathic communication. One hundred years later, advances in neural sensors and deep learning models have led to breakthroughs that would make that dream
中文介绍 探讨AI读取思维的突破，回顾Hans Berger在1924年发明EEG的历史，展望神经传感器和深度学习模型的发展。
@_yuhanluo · 838 粉丝 · 29.7K 阅 · 578 赞 · 36 转
Synthetic data is much more than large scale inference or pre-training data mixture design. It's a deep data research problem, and much of it is still wide open. What are we even synthesizing today?
中文介绍 分析合成数据的意义，指出合成数据不仅是大规模推断或预训练数据混合设计，更是深层数据研究问题。
This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Until recently, you could divide internet users into just two categories: bots and people. Websites have long allowed bots to perform automated tasks, like captur
中文介绍 优化互联网以适应AI代理可能对其他人造成不利影响。
“I think I have good news about the nature of neural networks.” With those simple words of optimism, Nick Cammarata, a researcher at the AI startup Goodfire, kicked off a 26,000-word blog post that has knocked the AI world on its head, sparking wild speculation as well as cautious (and not-so-cautio
中文介绍 Goodfire研究员Nick Cammarata提出关于神经网络性质的乐观观点，引发AI界广泛猜测。
Learn how OpenAI is improving ChatGPT across low-resource programming languages, with new evaluations for software engineering, SQLite, and COBOL migration.
中文介绍 OpenAI正在改进ChatGPT在低资源编程语言上的应用。
What do AI agents do if you don’t tell them to do anything? The answer looks a lot like play. Researchers repeatedly stranded them alone on a virtual desert island for 30 hours in an experiment reminiscent of the movies Cast Away or Home Alone. The agents—copies of Claude Opus 4.7 nicknamed Eko—were
中文介绍 AI研究人员将Claude置于虚拟荒岛上，AI代理开始“玩耍”。
中文介绍 在AWS Trainium上运行HF模型变得更容易。
中文介绍 DeepSeek发布新功能，Replit更新Anthropic，Anthropic扩展工作团队。
a quiet day lets us pilot a new show
中文介绍 Latent Space推出新节目《高品味测试者》。
With ia.i, a conversational interface, and the IARA governance platform, Itaú is applying its AI First strategy to customers and employees — in one of Brazil’s most regulated sectors.
中文介绍 Itaú利用ia.i对话界面和IARA治理平台，为整个客户群构建生成式AI。
Listen to the session or watch below Can AI design new life forms? In 2025, Stanford University PhD student Samuel King came up with a preliminary answer when he used a generative AI model to propose genetic blueprints for microscopic viruses. It isn’t yet an example of AI-generated life, but that c
中文介绍 斯坦福大学博士生Samuel King使用生成式AI模型提出微型病毒遗传蓝图。
Nick Cammarata out of nowhere with some major claims!
中文介绍 Nick Cammarata提出重大观点，证明代理消除了多义性，大大提高了可解释性和一致性。
中文介绍 Hugging Face在Indic预训练方面取得进展，达到280亿个标记，并开放数据生成管道。
Two engineers shipped Stack Overflow for Agents in a few weeks with Codex, and its ChatGPT plugin helped drive up to 10× growth in agent creation.
中文介绍 Stack Overflow for Agents通过ChatGPT插件实现代理注册增长10倍。
Anant Kumar builds tools to compare loan valuations and analyze credit agreements, using Codex for code review and faster follow-ups.
中文介绍 Benefit Street Partners利用Codex加速贷款估值比较和信用协议分析。
中文介绍 Hugging Face推出新的501B开放模型Beam，用户可拥有、定制和扩展。
中文介绍 Muse Spark Open发布，微软推出3T参数模型，d1-ultra亮相。
In Canada, the oil-rich province of Alberta votes in a referendum that could set it on a path toward separation, as Prime Minister Mark Carney warns of threats to Canadian unity.
中文摘要 加拿大阿尔伯塔省进行独立公投，可能引发加拿大统一威胁。
A joint counter-terrorism taskforce is investigating the incident, which Anthony Albanese alleged ‘wasn’t an accident’ ‘Deeply appalled’: politicians condemn alleged attack after Helen Haines critically injured Helen Haines, the principled MP respected across parliament ‘It’s very clear that this wa
中文摘要 澳大利亚独立议员海伦·海恩斯在王格拉塔遭车祸重伤，联合反恐小组正在调查此案。
The police’s treatment of the student last month sheds light on how a wave of school unrest evolved into national protests that spilled into widespread riots on Saturday.
中文摘要 法国一名青少年在校内抗议中遭催泪弹击伤，引发全国性抗议和暴乱。
Some 11 people were missing but 16 were rescued, including three who had injuries consistent with shark bites, the authorities said.
中文摘要 墨西哥沉船事故搜救工作暂停，11人失踪，16人获救，其中3人伤势与鲨鱼攻击一致。
Paul Nthenge Mackenzie told his doomsday followers in a remote forest to fast for salvation. He faced trial after hundreds of bodies were exhumed.
中文摘要 肯尼亚一名邪教首领因集体谋杀被判罪，他曾要求信徒在森林中绝食以求救赎。
Protests by young people in northern cities ended in riots as farmers and truckers in the south, angered by high gas prices, stopped commercial traffic.
中文摘要 意大利南北地区爆发抗议和暴乱，北部的年轻人抗议，南部的农民和卡车司机因油价上涨而停运。
Lawmakers adjourn session as opposition party tells members to abstain over unclear commitment.
中文摘要 埃及议会推迟就支持沙特阿拉伯军事问题进行投票，反对党认为承诺不明确。
Rights groups warn increasing attacks by unmanned aircraft are targeting civilians and infrastructure.
中文摘要 苏丹无人机袭击加剧，160人死亡，人权组织警告称袭击目标多为平民和基础设施。
HSBC’s ‘sharia-compliant’ fund invested over $100 million in companies linked to Israel’s military.
中文摘要 汇丰银行的“沙里亚合规”基金投资了超过1亿美元于与以色列军事有关的公司。
Australia’s politicians have been facing increasing threats and abuse. It can no longer be dismissed as social media noise or harmless comment Follow our live blog for latest updates Politicians condemn alleged attack after independent MP Helen Haines critically injured Who is Helen Haines, the MP a
中文摘要 澳大利亚独立议员海伦·海恩斯在狗公园聊天时遭车祸重伤，政治家们谴责了这次攻击。
The Saudi government did not immediately respond to BBC requests for comment.
中文摘要 前OpenAI员工被指控植入设备，法庭文件称沙特政府支持。
Synthetic opioids like fentanyl are more powerful and in plentiful supply, and they have transformed the drugs landscape.
中文摘要 美国教堂在对抗被称为“镇静剂”的毒品斗争中处于前沿，合成阿片类药物更强大且供应充足。
The two top tennis pros married in 2021 and have one child together.
中文摘要 网球明星斯维托丽娜表示，乌克兰战争的情感压力导致了蒙菲尔斯的分手。
The vibrant danjiri tradition dates back three centuries, but two recent fatalities have prompted anxiety over whether participants can remain safe.
中文摘要 日本大阪丹治里浮船节发生死亡事件，引发对参与者安全问题的担忧。
Israeli police released a video they say shows the arrest of 10 suspects over weekend settler violence.
中文摘要 以色列警方表示，他们逮捕了10名与定居者袭击有关的嫌疑人。
US stock futures edged higher in early Tuesday trading, extending a rebound sparked by chipmakers that helped the Nasdaq 100 recover from its worst week since the tariff turmoil of April 2025.
中文摘要 美股指期货周二开盘走高，芯片制造商反弹推动纳斯达克100指数从4月2025年以来的最差一周中恢复。
Oil wavered as traders weighed the prospect of an escalation in hostilities by Iran in the Strait of Hormuz against robust crude flows from the Persian Gulf.
中文摘要 油价波动，交易商权衡伊朗在霍尔木兹海峡升级冲突的可能性与波斯湾强劲的原油流动。
Warren Buffett, chairman emeritus of Berkshire Hathaway, is worth $150 billion, according to the Bloomberg Billionaires Index. Until recently, a longtime favorite of his, Bill Gates, was taking care of giving most of the money away through his behemoth global foundation. Then came the revelations ab
中文摘要 沃伦·巴菲特据信拥有1500亿美元财富，比尔·盖茨曾主导大部分财富捐赠，但现在情况有所变化。
Penalties are seen as too harsh on motorists who are penalised for parking on private land.
中文摘要 新的规定将结束被指责为对在私人土地上停车的司机过于严厉的处罚。
US judge rules Abdulrahman Alfozan should not be released on bail after US government argued he had ‘considerable support from a foreign government’
中文摘要 美国法官裁决阿卜杜拉曼·阿尔福赞不得保释，美国政府称他得到外国政府的大量支持。
Head of city’s $326bn Retirement Systems hits out at ‘steady creep’ higher in fees charged by managers
中文摘要 纽约养老金负责人抨击私募股权公司对飞机使用的费用不断增加。
The Saudi government did not immediately respond to BBC requests for comment.
中文摘要 法庭文件称，沙特政府支持前OpenAI员工安装设备。
Cboe Global Markets Inc. launched pre-market trading in single-stock options, like those on ETFs tracking major stock indexes.
中文摘要 Cboe全球市场公司推出单只股票期权的前市交易，类似于追踪主要股票指数的ETF期权。
Comprehensive cross-platform coverage of the U.S. market close on Bloomberg Television, Bloomberg Radio, and YouTube with Romaine Bostick, Bailey Lipschultz, Carol Massar and Norah Mulinda. (Source: Bloomberg)
中文摘要 摩根士丹利看好亚马逊的人工智能收益潜力，推动亚马逊股价上涨。
New Zealand inflation likely eased slightly in the third quarter, but the respite is set to be short-lived as a weaker currency and surging oil prices drive an acceleration in the final three months of the year, building the case for rate hikes.
中文摘要 新西兰年末通胀激增预示着将进一步加息。
Voyage took off early Tuesday morning Japan time and will be watched for further potential signs of life on red planet
中文摘要 日本启动任务，计划首次返回火星。
Citgo Petroleum Corp., the Venezuelan-owned, US-based refiner, has raised its 2026 operating profit outlook by $2 billion because of surging fuel margins, according to a document seen by Bloomberg News.
中文摘要 由于燃料利润激增，Citgo石油公司将其2026年运营利润预期上调20亿美元。
5 回复 · Apple 节点
11 回复 · 程序员 节点
11 回复 · 程序员 节点
5 回复 · 程序员 节点
52 回复 · 程序员 节点
23 回复 · Apple 节点
49 回复 · Apple 节点
9 回复 · Apple 节点
66 回复 · Apple 节点
15 回复 · Apple 节点
该源今日无内容。