📡 外网新声
译文为机器翻译，供速览；点开原文为准。外链观点不代表本站立场。
- Hacker News刚收ENContext Anxiety: How AI Agents Panic About Their Perceived Context Windowsagent
- Hacker News刚收ENTinyContext Guide – How to Add Long-Term Memory to an AI Agent (Step by Step)agentmemory
- Hacker News刚收ENWhy AI agents struggle with chip design (RTL) debugging, and what helpsagentchip
- Hacker News刚收ENShow HN: Zero-knowledge zero-overhead LLM routing mesh using pure web primitivesllm
- Hacker NewsENShow HN: AgentTM – A time-travel debugger for AI coding agents built in Goagent
- Hacker NewsENAn Illustrated Guide to AI Agentsagent
- Hacker NewsENReliably aligning LLMs is impossible in principlellm
- Hacker NewsENCEPO: General Framework for Improving Prompt Optimization in Multi-Agent Systemsagentmulti-agent
- Hacker NewsENShow HN: 13M opinions on 48k products extracted from Reddit and HNllm
- Hacker News译Show HN: Arken –调用MCP服务器/API的AI 智能体的加密身份agentcryptographShow HN: Arken – cryptographic identity for AI agents calling MCP servers/APIs
- GitHub 新星精译earendil-works/pi：智能体工具箱（统一 LLM 接口·循环·终端界面·编码 CLI）agentllmearendil-works/pi — AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI
- GitHub 新星精译ggml-org/llama.cpp：C/C++ 大模型推理引擎llmggml-org/llama.cpp — LLM inference in C/C++
- GitHub 新星精译NousResearch/hermes-agent：与你一起成长的智能体agentNousResearch/hermes-agent — The agent that grows with you
- GitHub 新星精译anomalyco/opencode：开源编码智能体agentopen sourceanomalyco/opencode — The open source coding agent.
- GitHub 新星译morluto/rea —使用代理对任何内容进行反向工程，从应用行为到本机二进制文件。agentmorluto/rea — Reverse engineer anything with agents, from app behavior down to native binaries.
- MIT 科技评论ENThe agentic web is hereagentagentic
- GitHub 新星译IBAX-io/go-ibax —一个创新的区块链协议平台，每个人都可以部署自己的应用程序daoblockchainIBAX-io/go-ibax — An innovative Blockchain Protocol Platform, which everyone can deploy their ow
- Hacker News译Show HN：EchoChain，了解LLM播放电话时会发生什么llmShow HN: EchoChain to see what happens when LLMs play telephone
- GitHub 新星译DAO-AILab/flash-attention -快速且内存高效的精确注意memorydaoDao-AILab/flash-attention — Fast and memory-efficient exact attention
- GitHub 新星译affaan-m/ECC —代理利用性能优化系统。技能、本能、记忆、安全、agentmemoryaffaan-m/ECC — The agent harness performance optimization system. Skills, instincts, memory, sec
- GitHub 新星译mattpocock/skills —真正工程师的技能。直接来自我的.agents目录。agentmattpocock/skills — Skills for Real Engineers. Straight from my .agents directory.
- Hacker News译YC的Minecraft和Mindustry agent swarms展示了多agent团队agentmulti-agentWhat YC's Minecraft and Mindustry agent swarms show about multi-agent teams
- arXiv cs.LG译统一多式联运模型连续空间中的潜在推理reasoningragLatent Reasoning in Continuous Space for Unified Multimodal Models
- arXiv cs.AI译潜在稀疏注意力：向次二次超长语境注意力迈进agentagenticLatent Sparse Attention: Towards Subquadratic Ultra-Long Context Attention
- arXiv cs.AI译具有隐式动态的层次世界模型agentplanningHierarchical World Models with Implicit Dynamics
- arXiv cs.AI译RoboArchitect：通过持久记忆实现机器人智能体的自动化设计agentmemoryRoboArchitect: Automating the Design of Robotic Agents with Persistent Memory
- arXiv cs.LG译两个ReLU神经元的梯度下降：全局景观和分岔动态neuralGradient Descent on Two ReLU Neurons: Global Landscape and Bifurcation Dynamics
- arXiv cs.AI译ACCT：非对称跨LLM上下文传输llmlarge languageACCT: Asymmetric Cross-LLM Context Transfer
- arXiv cs.AI译使用分层树结构化索引器进行次平方注意llmlarge languageSubquadratic Attention with a Hierarchical Tree-Structured Indexer
- arXiv cs.LG译物理知情神经网络（ Burgers ）中的硬约束投影neuralHard Constraint Projection in a Physics Informed Neural Network (Burgers)
- arXiv cs.LG译双曲逆强化学习computereinforcementHyperbolic Inverse Reinforcement Learning
- arXiv cs.LG译了解可塑性损失对随机环境中强化学习的影响reinforcementTowards Understanding the Impact of Plasticity Loss on Reinforcement Learning in Stochastic Envi
- arXiv cs.LG译RAMA：用于LEO卫星网络的具有情景记忆的检索增强型多Agent路由agentmulti-agentRAMA: Retrieval-Augmented Multi-Agent Routing with Episodic Memory for LEO Satellite Networks
- arXiv cs.AI译CRoS：加扰CoT的因果奖励提高了准确性和推理忠诚度reasoningreinforcementCRoS: Causality Reward on Scrambled CoTs Improves Accuracy and Reasoning Faithfulness
- arXiv cs.LG译来自物种级检测的丰富度敏感性声学生物多样性指数：分类器基准测试、假阳性控制和对标准声景指数的验证ibenchmarkenergyA richness-sensitive Acoustic Biodiversity Index from species-level detections: classifier bench
- arXiv cs.AI译相同的问题，不同的答案：审核六位LLM戒烟建议的人格依赖性差异llmlarge languageSame Question, Different Answers: Auditing Persona-Dependent Variation in Six LLMs on Smoking-Ce
- arXiv cs.AI译从噪声测量中学习清洁生成过程diffusionLearning Clean Generative Processes from Noisy Measurements
- arXiv cs.AI译没有自由循环：循环训练模型何时为其推理计算付费？memorycomputeThere Is No Free Loop: When Does Looping a Trained Model Pay for Its Inference Compute?
- arXiv cs.LG译量子核分类方法的实证评估：与经典核的比较研究evaluationbenchmarkEmpirical Evaluation of Quantum Kernel Methods for Classification: A Comparative Study with Clas
- arXiv cs.AIENKnow When the Equation Is Wrong: Structural Uncertainty for LLM-Guided Symbolic Regression under Extrapolationllmlarge language
- arXiv cs.AIENTradeWar: A Benchmark for Execution-Grounded Trade Negotiation with Adaptive LLM Agentsagentllm
- arXiv cs.AIENSecPO: Principled Adversarial Training for Prompt Injection Securityagentllm
- arXiv cs.AIENDAMI: Dual-Axis Memory Initialization for Enhanced Long-Context Retrieval in Gated Linear Attentionmemoryretrieval
- arXiv cs.AIENID2Pattern: Pretraining Generalizable Identity Behavior Patterns for Transaction Risk Foundation Modelspayment
- arXiv cs.AIENPrune Gradually, Adapt Sparsely: Efficient Pruning of Large Language Models at Extreme Sparsityllmlarge language
- Hacker News译Show HN：HiveOS -具有PII药丸和3型号共识的多代理企业操作系统agentmulti-agentShow HN: HiveOS – Multi-agent enterprise OS with PII pills and 3-model consensus
- DeepMind 官方ENSpeeding up planning decisions with AIplanning
- OpenAI 官方译社区安全报告和针对青少年的新保护措施safetyCommunity safety report and new protections for teens
- GitHub 新星精译rasbt/LLMs-from-scratch：一步步用 PyTorch 从零实现类 ChatGPT 模型llmrasbt/LLMs-from-scratch — Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
- Hacker News译问HN：Agent swarm，orchestra，sequential，你的工作流程是什么？agentmulti-agentAsk HN: Agent swarm, orchestra, sequential, what's your workflow?
- GitHub 新星译obra/superpowers —有效的代理技能框架和软件开发方法。agentagenticobra/superpowers — An agentic skills framework & software development methodology that works.
- MIT 科技评论译我们迫切需要实时监控AI 智能体的行为agentWe urgently need to monitor AI agents’ actions in real time
- OpenAI 官方译我们如何检测前沿模型错位agentalignmentHow we detect frontier model misalignment
- MIT 科技评论译利用企业环境创造人工智能价值automationCreating AI value with enterprise context
- OpenAI 官方译OpenAI和ServiceNow扩展跨企业工作流程的合作伙伴关系agentOpenAI and ServiceNow expand partnership across enterprise workflows
- OpenAI 官方译Legora正在为法律工作构建代理操作系统agentagenticLegora is building the agentic operating system for legal work
- Hacker News译在GCP上使用Multi-Agent Swarms构建对话式商务agentmulti-agentBuilding Conversational Commerce with Multi-Agent Swarms on GCP
- OpenAI 官方译Retell使用OpenAI模型来帮助推动每月1亿次通话agentbenchmarkRetell uses OpenAI models to help power 100M monthly calls
- GitHub 新星译TauricResearch/TradingAgents — TradingAgents：多代理LLM金融交易框架agentmulti-agentTauricResearch/TradingAgents — TradingAgents: Multi-Agents LLM Financial Trading Framework
- OpenAI 官方译OpenAI扩大与甲骨文的合作伙伴关系，将其应用程序引入ChatGPTagentOpenAI expands partnership with Oracle to bring its apps to ChatGPT