Python · ★ 52,966 · 🍴 17,333 · 📈 2,213 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 职业搜索工具，基于 Claude Code，提供职位评估、简历定制、求职信撰写和面试准备等功能。
Python · ★ 42,638 · 🍴 4,526 · 📈 1,187 stars today
Official, Anthropic-managed directory of high quality Claude Code Plugins.
中文介绍 Anthropic 管理的 Claude Code 插件官方目录，提供高质量插件。
TypeScript · ★ 216,512 · 🍴 28,987 · 📈 1,417 stars today
The open source coding agent.
中文介绍 开源编码代理，用于自动化代码编写。
TypeScript · ★ 12,715 · 🍴 1,253 · 📈 351 stars today
Cursor plugin specification and official plugins
中文介绍 Cursor 插件规范和官方插件，提供插件开发和管理工具。
TypeScript · ★ 76,478 · 🍴 9,041 · 📈 502 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 智能多玩家集群部署工具，支持自主工作流协调和对话式 AI 系统，具备自适应记忆和自学习智能。
Rust · ★ 10,445 · 🍴 748 · 📈 513 stars today
Solution for long term memory for agent coding CLIs and to facilitate handoff between different agent vendors
中文介绍 为智能代理编码 CLI 提供长期记忆解决方案，并促进不同代理供应商之间的交接。
TypeScript · ★ 96,560 · 🍴 6,173 · 📈 1,146 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 Orca 平台支持运行多个代理，如 Grok Bot 和 Claude Code，可在桌面、移动和远程环境中使用。
Python · ★ 54,556 · 🍴 4,945 · 📈 582 stars today
"CLI-Anything: Making ALL Software Agent-Native" -- CLI-Hub: https://clianything.cc/
中文介绍 CLI-Anything 项目，旨在使所有软件都支持代理，提供 CLI-Hub 以方便插件管理。
Python · ★ 75,218 · 🍴 8,234 · 📈 2,098 stars today
VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages.
中文介绍 VoiceStudio 是一个开源的语音克隆、设计、配音、转录和有声书创作工具，支持多种语言。
Kotlin · ★ 8,743 · 🍴 567 · 📈 381 stars today
中文介绍 Android 技能项目，具体内容未提供。
★ 83,994 · 🍴 9,780 · 📈 81 stars today
Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.
中文介绍 大型语言模型（LLM）入门课程，提供路线图和 Colab 笔记本。
Python · ★ 2,368 · 🍴 184 · 📈 115 stars today
Video restoration tool
中文介绍 视频修复工具，用于改善视频质量。
Python · ★ 138,572 · 🍴 16,498 · 📈 155 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 功能强大且模块化的扩散模型 GUI、API 和后端，具有图形/节点界面，提供世界上最快的本地推理引擎。
👍 6
Modern generative models produce high-quality individual outputs but struggle to sample diversely, even when explicitly instructed to do so. We introduce MINE-SAMPLE-RENDER (MSR), a test-time approach that externalizes the diversity into an explicit, corpus-derived sampling procedure. Our pipeline (
👍 8
Computer-use agents increasingly tackle long-horizon tasks, where running a frontier model at every step is expensive and a small open-weight model alone often fails. Weak-to-strong escalation offers a middle ground where a weak agent acts by default and hands control to a strong one only when neede
👍 3
Personalization of Large Language Models (LLMs) is increasingly important for user-centric applications, yet prior methods mainly adapt dense LLMs, whose polysemantic neurons, each encoding multiple unrelated concepts, hinder localizing the components behind personalization, and thus understanding a
👍 2
Reusing the attention KV cache across requests is a standard way to cut prefill cost in long-context LLM serving, and production systems increasingly back the per-engine cache with a remote tier shared by every engine. We study such a tier in a disaggregated deployment of GLM-5.3, an FP8 mixture-of-
👍 8
Single image 3D generation aims to dress a given 3D mesh with relightable materials while faithfully preserving the visual details of a reference image. Existing methods either introduce artifacts when fusing inconsistent multi-view predictions or struggle to preserve fine details through implicit i
👍 285
Existing agent benchmarks typically evaluate agents on well-specified tasks defined by issue reports, specifications, or task tickets. Real-world customer delivery is different: forward-deployed engineers (FDEs) must clarify ambiguous requirements, work in live systems, and uncover unstated constrai
👍 28
Data agents use code to process heterogeneous files and derive analytical insights. Scaling their training faces costly environment curation, privacy and intellectual-property restrictions, and difficulty verifying open-ended findings without known generative truth. Our core insight is to specify th
👍 32
Generative visual verifiers play a crucial role in advancing multimodal intelligence. To enhance the robustness of visual verifiers in complex multimodal scenarios involving compositional requirements, we move beyond conventional binary verifiers that produce only True/False discrete judgments and i
👍 2
As Vision-Language Models (VLMs) advance, they are increasingly expected to perform spatiotemporal reasoning over motion, actions, and view changes beyond what the input images show. Video World Models (VWMs) provide a natural unified engine for visual imagination by generating task-relevant visual
👍 316
Computer use has reached a turning point: the focus is expanding from GUI-only interaction to systems that complete tasks on users' computers through visual and programmatic interfaces. We define a generalist computer-use agent (GCUA) as a system in which the model chooses how to observe, act, and o
👍 354
Spatial intelligence lets an agent perceive and act in the physical world, and many of its judgments must be made at interaction rates: which spatial relation holds, how far an object is, or where to place it. Vision-language models (VLMs) produce these answers by free-form text generation, paying f
👍 54
Vision-language-action (VLA) models transfer semantic knowledge acquired through vision-language pretraining to robot control, achieving strong performance across various benchmarks. However, most VLAs directly map observations to demonstrated actions without explicit supervision of intermediate phy
👍 668
Policy-shift distillation reuses the difference between a teacher's log probabilities before and after reinforcement learning (RL) as supervision on student-generated trajectories. However, this endpoint policy shift cannot reveal whether local action preferences evolved consistently during teacher
👍 5
Multimodal deep search is moving from short queries to long-horizon investigations that continually acquire evidence. However, training such agents remains constrained by limited challenging training questions and insufficient process feedback. We introduce MMSearch-OPD, combining controllable dat
👍 25
Large language models (LLMs) are expanding from assisting scientific research and engineering to autonomously developing and improving AI systems, advancing AI for AI (AI4AI) as a pathway toward recursive self-improvement (RSI). Workflow-driven machine learning engineering (MLE) systems provide a pr
👍 11
The capabilities of an agent swarm depend on both its individual members and how they coordinate. Yet existing multi-agent systems often train agents within fixed orchestration structures, separating the learning of individual capabilities from the organization of collective behavior. We introduce m
👍 2
Character Error Rate (CER), the standard metric for OCR and vision-language models (VLMs), charges a typo as much as a wrong name. We introduce semCER, which keeps CER's alignment and denominator and charges each grapheme edit a severity from an LLM judge, with no word segmentation. Normalized CER k
👍 3
When does a language model acquire a capability? Developmental studies usually answer this by tracking when models start producing the correct token. But that prediction passes through the unembedding matrix, or readout, which is itself still being learned during pretraining. We show that a model's
👍 154
Online reinforcement learning (RL) with verifiable rewards has driven rapid progress in mathematics, coding, and tool use, but scaling this success to GUI agents remains challenging due to the high cost of interactive environments and the scarcity of realistic training tasks. We introduce MAI-Gym, a
👍 5
Long-horizon agentic RL is constrained by the futures visited by the current policy. When rollouts fail to reach useful behaviors, improved credit assignment cannot provide supervision for states and actions that were never encountered. On-policy distillation (OPD) introduces teacher guidance on pol
👍 5
Transformers are intuitively considered to be efficient at copying from their inputs. We actually show that the ability of LLMs to copy in context can be improved through explicit modeling. We propose Copy-Paste Language Models (CPLMs), which add a lightweight Copy-Paste head to the conventional lan
👍 10
The design of attention mechanisms in long-context large language models faces an ``impossible triangle'' among KV cache size, training and prefill FLOPs, and model quality. GQA achieves strong quality but incurs high KV cache overhead; alternative approaches such as MLA and MFA can compress the cac
👍 8
Cowork agents navigate dependencies across tools and files while adapting to changing user requirements and edits to shared workspaces. Scaling such environments requires maintaining consistency among task objectives, executable resources, user behavior, and success criteria. We introduce Cowork-Gym
👍 3
Training-free diffusion-based compression methods repurpose the generative priors of a diffusion or a flow model to compress data without retraining them as task-specific codecs. Recent codebook-based approaches such as Turbo-DDCM achieve strong perceptual quality at ultra-low bitrates, but their re
👍 8
Reinforcement learning with verifiable rewards (RLVR) typically uses small rollout groups per prompt, where repeated sampling from the same policy can yield highly similar trajectories, reducing within-group reward variation and weakening group-relative advantage signals. Existing remedies such as h
👍 2
Generalist robot manipulation can benefit from pretrained vision-language models (VLMs) for semantic reasoning and video generation models (VGMs) for world modeling. However, existing unified models that directly fuse them through shared attention at multiple layers can make it harder to draw on the
👍 4
We study environment scaling as a way to increase agent-benchmark difficulty while holding core task objectives fixed. Our framework, Recursive Environment Scaling (RES), transforms existing benchmark environments while preserving their objectives and completion requirements. A proposer changes info
👍 4
Agent environments are essential for large language model (LLM) development. They use programmatic verifiers to turn complex outcomes into rewards. Verifiers can be attacked into awarding full credit to cheating agents without completing the intended task, causing inaccurate evaluation and misalignm
👍 3
Automatic speech recognition produces words but drops the non-verbal content of a recording: the breaths, coughs, sighs and laughter of the speaker, and the horns, music or barking around them. We describe Speech Event Detector, a 5.8-million-parameter convolutional-Transformer model that outputs pr
👍 1
Vision-language-action (VLA) models map visual observations and language instructions to continuous robot actions. This task requires a transition from representations that describe the scene and instruction to representations that support action generation. Many continuous-action VLAs leave this tr
@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 回顾百年前脑电波发明，探讨神经传感器与深度学习模型在心灵感应领域的突破。
@_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 深入分析合成数据在深度数据研究中的重要性，指出当前合成数据领域的开放性问题。
@arthurmensch · 88.8K 粉丝 · 17.9K 阅 · 619 赞 · 109 转
An idea has gained ground in Europe: we should leave frontier AI models to the Americans and the Chinese. It is advocated by serious people, whose reasoning has its logic. A frontier model costs
中文介绍 探讨欧洲关于将前沿AI模型留给美中两国的观点，分析其背后的逻辑与成本问题。
@BradGroux · 6.9K 粉丝 · 1.3K 阅 · 512 赞 · 3 转
On the drive home from the airport after OpenAI DevDay, I spent about 45 minutes on a phone call with AVA, my dot. Before I got home, I had planned the next day and moved work forward across several
中文介绍 分享OpenAI DevDay的收获，包括与AVA的通话体验和规划次日工作的进展。
a quiet day lets us pilot a new show
中文介绍 Latent Space推出新节目《High Taste Testers》的首播。
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 巴西银行Itaú推出面向客户和员工的生成式AI产品ia.i和IARA治理平台，应用AI First策略。
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设计微生物病毒基因蓝图，探讨AI设计新生命形式的可能性。
Nick Cammarata out of nowhere with some major claims!
中文介绍 Latent Space的Nick Cammarata提出重大声明，旨在提高AI的可解释性和一致性。
中文介绍 Hugging Face Blog报道了NVIDIA的Indic预训练规模扩大至2.8万亿个标记和开放数据生成管道。
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进行代码审查和更快跟进，提高贷款估值和信用协议分析效率。
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插件增加注册量，驱动代理创建增长10倍。
中文介绍 Hugging Face推出Beam模型，一个用户可拥有、定制和扩展的501B开源模型。
中文介绍 TLDR AI报道了微软的3T参数模型d1-ultra和Muse Spark Open的开放。
中文介绍 Hugging Face Blog介绍了IBM的Granite Switch 4.1，提供聊天、决策和语音功能。
Introducing the Rosalind Academic Plan, offering eligible academic labs sponsored access to GPT-Rosalind and tools to advance life sciences research.
中文介绍 OpenAI推出Rosalind学术计划，为符合条件的学术实验室提供GPT-Rosalind和生命科学研究工具的赞助访问。
Accelerating the search for novel antimicrobial peptides using Google Antigravity for Science.
中文介绍 DeepMind使用Google Antigravity加速新型抗菌肽的搜索。
Celebrate Teachers’ Day with AI for Educators in Brazilian Portuguese and see how Study Mode can help teachers support active, step-by-step learning.
中文介绍 OpenAI推出AI教育产品，庆祝教师节，并介绍如何通过问题学习模式支持学生积极学习。
Sandboxes are all you need.
中文介绍 Modal实现10亿美元年度经常性收入和160亿美元D轮融资，成为纽约市最新的AI基础设施独角兽公司。
In collaboration with OpenAI, Sullivan & Cromwell developed Agreement Analyzer, an AI tool for identifying issues in a draft agreement.
中文介绍 Sullivan & Cromwell与OpenAI合作开发Agreement Analyzer，用于识别草稿协议中的问题。
After anti-government protests devolved into chaos, leaders from across the political spectrum blamed one another for the unrest, even as it largely subsided on Sunday.
中文摘要 法国政府抗议活动升级为暴乱，多处建筑和商店受损，政治领导人相互指责。
The exiled Nobel laureate has petitioned for new elections in Venezuela and for the country to allow her return.
中文摘要 委内瑞拉反对派再次支持流亡诺贝尔奖得主玛丽亚·科琳娜·马查多担任领导人。
Prime minister Mark Carney describes blaze as ‘one of the worst tragedies for the forces on Canadian soil in peacetime’ Four people have been killed and dozens more seriously injured after a “violent, fast-moving” fire tore through one of Canada’s busiest military training bases in Alberta, Canada’s
中文摘要 加拿大军事基地发生火灾，四人遇难，数十人受伤，加拿大总理马克·卡尼称之为和平时期加拿大军队最严重的悲剧。
The US president blocked numerous sanctions on Russia two days before they were due to take effect.
中文摘要 特朗普推迟对俄制裁以维持和平谈判。
Alberta’s independence movement faces likely defeat, but history suggests losing a referendum may not end the fight.
中文摘要 加拿大艾伯塔省独立公投面临失败，但历史表明失败未必意味着斗争结束。
Students upset by living and educational costs have revived protests. Prime Minister Edi Rama said they were being used by “outside forces.”
中文摘要 阿尔巴尼亚总理埃迪·拉马谴责暴力抗议，称抗议者被“外部力量”利用。
At least four people are confirmed to have died after a huge fire swept through the base.
中文摘要 加拿大军事基地火灾已确认四人遇难。
Police responding to reports of an alleged stolen car in Dulwich Hill were told a man was armed with a firearm Get our breaking news email, free app or daily news podcast A man has been charged with a slew of offences following an investigation into reports a man was allegedly seen with a firearm in
中文摘要 悉尼男子因涉嫌枪击事件被捕，导致Dulwich Hill地区封锁。
French President Emmanuel Macron says protesters who attacked government buildings need to pay for their crimes.
中文摘要 法国总统马克龙谴责暴乱导致的抢劫和大规模逮捕事件。
The BBC's Quentin Sommerville reports from Chicago, where communities are battling a new wave of synthetic drugs.
中文摘要 芝加哥社区对抗涌入美国的合成药物新潮流。
The BBC's Quentin Sommerville reports from Chicago, where communities are battling a new wave of synthetic drugs.
中文摘要 芝加哥社区对抗涌入美国的合成药物新潮流。
Pakistan's political showdown continues as more opposition figures are implicated under Pakistan's anti-terror law.
中文摘要 巴基斯坦政治对峙持续，更多反对派人物被涉嫌违反反恐法。
The UN's World Food Programme says 1.4 million people in Gaza are suffering acute food shortages and that funding gaps are forcing cutbacks in critical aid, despite the U.S. brokered Hamas-Israel ceasefire.
中文摘要 世界粮食计划署负责人表示加沙地带生活“绝对残酷”，140万人遭受严重粮食短缺。
Prime Minister Mark Carney said the “violent, fast-moving” fire on Saturday in the province of Alberta was one of the worst tragedies for Canada’s armed forces in peacetime on Canadian soil.
中文摘要 加拿大军队基地火灾导致四人死亡，数十人受伤，加拿大总理马克·卡尼称之为和平时期加拿大军队最严重的悲剧。
The Amazon rainforest has been a crucial battleground in the country’s presidential runoff.
中文摘要 亚马逊雨林成为巴西总统大选的关键战场。
EQT AB has agreed to buy Australia’s Cleanaway Waste Management Ltd. through a unit, in a deal that values the firm at about A$9.3 billion ($6.5 billion) and makes it the latest local company to enter private hands after a period of share underperformance.
中文摘要 EQT同意以65亿美元收购澳大利亚Cleanaway Waste Management Ltd.，使该公司估值约为93亿澳元。
US stock futures rebounded from Friday’s technology-led selloff and oil slipped as investors turned to a packed week of corporate earnings and Chinese economic data for direction.
中文摘要 美股期货从上周五科技股抛售中反弹，油价下滑，投资者关注企业财报和中国经济数据。
Oil steadied after jumping on Friday, as flareups in fighting in the Middle East and Ukraine dimmed the prospect of a return to normal crude flows.
中文摘要 油价在周五上涨后稳定，中东和乌克兰冲突加剧降低了原油恢复正常流动的前景。
Daiichi Life Group Inc. offered to buy the remainder of Challenger Ltd. that it doesn’t already own, in a takeover bid that valued the Australian firm at around A$7.9 billion ($5.5 billion).
中文摘要 Daiichi Life提出44亿美元收购澳大利亚Challenger Ltd.剩余股份，该公司估值约79亿澳元。
Warren Buffett is worth an estimated $150 billion. He's also promised to divest himself of his vast wealth before he dies. Except it's hard when the pot keeps growing. So his three children are stepping in to spread it around. Alexandre Rajbhandari explains (Source: Bloomberg)
中文摘要 沃伦·巴菲特估值1500亿美元，承诺在生前处理其巨额财富，其子女将协助分配。
Kioxia Holdings Corp. shares face further downward pressure because of a buildup of positions from investors who borrowed to buy the stock on the assumption it would rise, and now may need to sell at a loss.
中文摘要 Kioxia股价因投资者融资买入后股价下跌而面临进一步下行压力。
Nigel Dawn and Naveen Nataraj are elevated while John Weinberg will remain chief executive for at least three years
中文摘要 Evercore首次任命联合总裁，John Weinberg将至少担任首席执行官三年。
Since Donald Trump came back into office, copper has surged on a seismic reshuffling of global inventories linked to the risk that he’ll impose tariffs on the metal. But as Trump nears the halfway mark on his second term, the market’s many bulls are growing wary that prices could crash if he doesn’t
中文摘要 由于没有迹象表明将征收关税，交易者对铜牛市日益谨慎。
US Treasuries are showing tentative signs of reclaiming their traditional role as a haven — threatening to wrong-foot investors who are heavily positioned for this year’s bond slump to deepen.
中文摘要 随着AI交易波动，债券市场的大规模空头押注受到警示。
The New Zealand dollar has fallen for five straight quarters and if analysts are right, it may be on course to test a pandemic-era low.
中文摘要 新西兰元连续五个季度下跌，可能测试疫情时期的低点。
US tech giant abandons agreement to lease capacity at neocloud’s UK site
中文摘要 微软取消与英国neocloud数据中心租赁协议。
Ambassador Andrew Puzder warns about ‘devastating’ retaliation if Brussels goes ahead with plans to limit spectrum access for non-EU companies
中文摘要 美国威胁对欧洲实施报复，若布鲁塞尔继续限制非欧盟公司频谱接入计划。
8 回复 · 程序员 节点
13 回复 · 程序员 节点
5 回复 · 程序员 节点
6 回复 · 程序员 节点
9 回复 · 程序员 节点
17 回复 · 程序员 节点
44 回复 · Apple 节点
8 回复 · Apple 节点
40 回复 · Apple 节点
8 回复 · Apple 节点
该源今日无内容。