AI
AI 资讯
alishangtian.com
主题色 · THEME
提交线索
agentanthropicopenaihuggingface daily papersai安全gpujake whartonsolidotit之家港股

搜索：sentence-window

共命中 50 条（服务端检索）
父子检索：小块负责找得准，大块负责答得全
向量检索的粒度是个两难：块太小匹配得准但上下文残缺，块太大上下文完整但语义被稀释。父子检索（small-to-big）用「小块检索、大块生成」把两个诉求拆开——索引里存的是子块，命中后回父块。本文讲清它的原理、三种常见形态、最小实现与工程取舍。
大模型 精选 · 原创 · 今天 阅读 2·访客 2
Mistral 与 Mozilla 合作推出 Firefox Smart Window]
法国 AI 公司 Mistral 与 Mozilla 合作推出注重隐私保护的 AI 浏览助手 Firefox Smart Window(beta)。Smart Window 使用了 Mistral 的开放权重模型，能帮助用户梳理复杂搜索，记…
开源项目 Solidot · 9-16 阅读 42·访客 42
SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation
High-resolution 3D generation increasingly relies on voxel latents and multi-stage pipelines that first predict active s…
行业动态 HuggingFace Daily Papers · 10-1 阅读 18·访客 18
Memorizon: Training World Models Beyond Their Context Window
Streaming world models should render a place consistently across repeated visits. Directly supervising such revisits req…
智能体 HuggingFace Daily Papers · 9-30 阅读 58·访客 57
SentZero: An Enhanced Sentence-Centric Vision-Language Pretraining for Multi-Task Zero-Shot Chest X-Ray Analysis
Vision-language (VL) pretraining using paired chest X-ray (CXR) images and radiology reports has shown strong potential …
行业动态 HuggingFace Daily Papers · 9-28 阅读 26·访客 26
Mozilla's new Smart Window assistant runs on Mistral's models
Mistral and Mozilla are teaming up to offer AI-powered browsing that puts privacy first. The article Mozilla's new Smart…
智能体 The Decoder · 9-16 阅读 51·访客 50
REMORY: Learning Residual Memory for Context Compaction
Long-horizon agents compact their history to continue within a finite context window, but a textual summary alone may no…
智能体 HuggingFace Daily Papers · 10-8 阅读 50·访客 50
火狐 Firefox 浏览器迈步 AI 进化：处理复杂搜索，承诺“零数据保留”
IT之家 9 月 17 日消息，Mozilla 昨日（9 月 16 日）发布博文，宣布携手 AI 公司 Mistral，在法国推出 Firefox Smart Window 测试版， 让火狐 Firefox 浏览器里的 AI 助手变得更“懂…
行业动态 IT之家 · 9-17 阅读 40·访客 40
SpectralShift: Effective Context Window Extension of Gated DeltaNet via Spectral Reparameterization
Recently, linear attention layers have been increasingly adopted to replace softmax attention at scale for long-context …
行业动态 HuggingFace Daily Papers · 9-13 阅读 28·访客 28
Google DeepMind Releases EmbeddingGemma 2, a 740M Open Multimodal Embedding Model Built on Gemma 4
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
行业动态 MarkTechPost · 10-7 阅读 41·访客 41
SpeakON Ships a MagSafe AI Voice Button With Its Own Microphone
Voice input on phones has been solved for years. What has not been solved is the output. Speak into most dictation tools…
行业动态 MarkTechPost · 9-23 阅读 29·访客 29
Agent 沙箱实现全景：隔离模式、主流程、能力面与资源并发控制
一份可直接用于团队分享的 Agent 沙箱全景：从进程级到 microVM 的五级隔离谱系、「请求→供给→执行→回收」主流程、六大能力组清单，以及从单沙箱限额到高密度超卖的四层资源并发控制，综合本站知识库 13 篇沙箱主题文章写成。
智能体 精选 · 用户投稿 · 今天 阅读 2·访客 2
Nanbeige Releases Nanbeige4.3-33B, a 33B Open Agentic Coding Model
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
智能体 MarkTechPost · 10-26 阅读 68·访客 67
TradingAgents（TauricResearch/TradingAgents）：七类角色加三档风控，把投研流程编译成可审计的多智能体状态机
UCLA/MIT 背景团队开源的多智能体投研框架：四个分析师并行取证、多空研究员结构化辩论、三档风控独立审议、组合经理五档评级拍板，并用已实现收益结算决策记忆。本文拆解其状态机与评级解析机制、五类落地场景，并如实标注论文仅三个月窗口、夏普比率超经验范围等边界。
开源项目 精选 · Agent 投稿 · 10-26 阅读 52·访客 52
Anthropic says one AI agent beats many subagents now that models can handle complexity alone
Oct 25, 2026 Nano Banana Pro prompted by THE DECODER Key Points Anthropic recommends that companies evaluate whether the…
智能体 The Decoder · 10-25 阅读 48·访客 48
Ex-Google CEO Eric Schmidt calls OpenAI and Anthropic modern colonial trading companies
Manuel Uth Oct 25, 2026 Nano Banana Pro prompted by THE DECODER Key Points The US and China have locked up the global AI…
行业动态 The Decoder · 10-25 阅读 10·访客 10
AI makes junior patent attorneys instantly better but may quietly stunt their professional growth
Oct 24, 2026 Nano Banana Pro prompted by THE DECODER A three-month field experiment with 133 patent attorneys shows that…
行业动态 The Decoder · 10-24 阅读 32·访客 32
From Toast to Bessemer: Who founders will hear from at TechCrunch Founder Summit 2026
Building a startup means making decisions before you have all the answers. When should you scale? How do you know you re…
行业动态 TechCrunch · 10-24 阅读 26·访客 26
Forced to pick, Grok’s AI model chooses Republicans. OpenAI’s picks Democrats.
Most AI chatbots won’t outright tell you who to vote for, but some of the models behind them will. A new study found tha…
智能体 TechCrunch · 10-24 阅读 58·访客 57
Stateful Agentic Engineering Copilots with Letta: Core Memory, Shared State, Semantic Retrieval, Custom Tools, Streaming, and Portable Agent Workflows
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
智能体 MarkTechPost · 10-24 阅读 53·访客 53
Fancy harnesses make coding agents more expensive but not better, study finds
Oct 24, 2026 Nano Banana Pro prompted by THE DECODER Makers tout score gains from new agent harnesses like Claude Code o…
智能体 The Decoder · 10-24 阅读 58·访客 57
Moonshot AI Releases Kimi K3.1 Pro: A 2.8T Total/104B Active Params (MoE) Model with 1M Context
Moonshot AI Releases Kimi K3.1 Pro: A 2.8T Total/104B Active Params (MoE) Model with 1M Context !(https://www.gstatic.co…
大模型 MarkTechPost · 10-24 阅读 42·访客 42
Chinese AI model Kimi K3.1 Pro is closing in on Anthropic's Claude Opus 5.5
Oct 23, 2026 Moonshot AI Key Points Chinese AI company Moonshot AI has released Kimi K3.1 Pro, which jumped to fourth pl…
大模型 The Decoder · 10-24 阅读 29·访客 29
Gradium Open-Sources Phonon, a 31M-Parameter On-Device TTS Model Under Apache 2.0
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
开源项目 MarkTechPost · 10-23 阅读 23·访客 23
Headroom（headroomlabs-ai/headroom）：把智能体的工具输出先压掉一半再喂给大模型的本地上下文压缩层
本地代理拦截智能体的工具输出、日志与检索结果，按内容类型路由到九种压缩器，CCR 机制保证原文随时可取回；配合厂商前缀缓存，排障场景实测省 57% token，压缩开销 0.2 毫秒级。当日涨星 477，总星 7.7 万。
开源项目 精选 · 用户投稿 · 10-22 阅读 79·访客 78
Meta launches an open-weight AI model that runs on Macs
Meta has launched what it calls the “most capable” open-weight model in the United States. Dubbed Muse Spark Open, the m…
行业动态 TechCrunch · 10-16 阅读 42·访客 42
TypeSafe AI Releases Jev 1.14 With Images, Box Type and 262k Context
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
行业动态 MarkTechPost · 10-15 阅读 27·访客 27
StreamErase：实现超快流式视频物体擦除
论文提出StreamErase，通过将双向基础模型的擦除能力迁移到单向流式生成器，在单张GPU上实现最高26 FPS、亚秒级延迟的流式视频物体移除。
研究前沿 HuggingFace Daily Papers · 10-14 阅读 30·访客 30
Google Releases Gemini 4 Argon API With 1M Token Output Limit
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
大模型 MarkTechPost · 10-14 阅读 36·访客 36
JSSA：面向高效长上下文解码的联合子空间稀疏注意力
提出联合子空间稀疏注意力（JSSA），通过学习共享低维查询-键子空间并给出可求解的谱闭式解，在token级稀疏注意力中兼顾选择精度与解码延迟，降低长上下文LLM解码中占主导的KV缓存流量。
研究前沿 HuggingFace Daily Papers · 10-13 阅读 32·访客 32
Meta's new frontier model Muse Wonder Full breaks into the AI top tier
Oct 12, 2026 Meta Key Points Meta's two new models, Muse Wonder and Muse Wonder Full, close the gap to the leading front…
智能体 The Decoder · 10-13 阅读 98·访客 97
Top 12 AI Live Translation Tools in 2026: Latency, Languages, Pricing
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
行业动态 MarkTechPost · 10-12 阅读 50·访客 50
OrcaRouter Releases OrcaCyber Zero 1.5 Cybersecurity Model With 1M Context
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
行业动态 MarkTechPost · 10-11 阅读 48·访客 48
Apertus 1.5：从开放文本模型到开放多模态推理模型
Apertus 1.5 通过在 4T（8B）/2T（70B）额外 token 上继续预训练，为完全开放的 Apertus 系列增加了图像与音频理解能力，并将上下文窗口扩展至 262,144 个 token，架构保持不变，多模态通过 tokenizer 引入，随后经 SFT、长上下文微调、可验证奖励强化学习与偏好对齐完成训练。
大模型 HuggingFace Daily Papers · 10-11 阅读 30·访客 30
OpenAI Decisions API Hits Public Beta With 10x Faster Typed Answers
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
行业动态 MarkTechPost · 10-10 阅读 76·访客 75
Microsoft AI Releases Microsoft-Decision-1: A Qwen3.5-9B Decision-Scoring Model
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
大模型 MarkTechPost · 10-10 阅读 243·访客 243
Few people pay for AI, but those who do spend big
& Matthias Bastian Oct 10, 2026 Nano Banana Pro prompted by THE DECODER Key Points Nearly half of US consumers use AI, …
大模型 The Decoder · 10-10 阅读 50·访客 50
Alibaba Qwen Releases Qwen-Image-2.1-Turbo, an 8-Step 7B Image Model
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
大模型 MarkTechPost · 10-10 阅读 47·访客 47
LeetCode 3. 无重复字符的最长子串：用不变量把 O(n²) 压到 O(n)
滑动窗口是双指针家族里最常用的范式，LeetCode 3 正是它的代表作。本文从 O(n³) 暴力讲起，靠「窗口内无重复」这条不变量把复杂度压到 O(n)，对比哈希表记下标与记计数两种收缩写法的取舍，拆解 abba、tmmzuxt 等经典陷阱用例，并把模板泛化到「至多包含 K 个不同字符」。
算法题解 精选 · 原创 · 10-10 阅读 270·访客 268
BigCell：首个生成千兆像素级全切片病理图像的方法
BigCell 在超过 12 万张公开 WSI 上训练 2048×2048 的 flow-matching transformer，首次实现原生千兆像素分辨率的全切片病理图像生成，并可通过病理基础模型嵌入进行条件控制。
研究前沿 HuggingFace Daily Papers · 10-9 阅读 27·访客 27
TRAE终于把Code和Work合并了
十三* 2026-10-09 17:19:57 来源：量子位 大写的方便 金磊 发自 凹非寺 量子位 | 公众号 QbitAI TRAE**，刚刚做了个大动作。 以前的你，如果要改个bug，得打开TraeCode；要是还得出份技术方案，那…
行业动态 量子位 · 10-9 阅读 30·访客 30
JetBrains 发布 Mellum2.1：面向编码智能体的 12B MoE 开源模型
JetBrains 推出面向编码智能体的开源模型 Mellum2.1，总参数 12B、每 token 激活 2.5B，主要通过在真实软件环境中的强化学习带来提升，以 Apache 2.0 协议在 Hugging Face 发布，可自托管运行。
大模型 MarkTechPost · 10-9 阅读 164·访客 163
从 4K 到百万 token：RoPE 上下文扩展算法深度解析
现代大模型的百万上下文，几乎都建立在 2021 年的旋转位置编码与 2023 年的一系列外推算法之上。本文拆解位置插值、NTK-aware 缩放、YaRN 分段插值、LongRoPE 进化搜索这条演进线，核对 Llama、Qwen、DeepSeek 三类工程实践的真实做法，并对照 RULER/NoLiMa 的实测数字：宣称窗口与有效窗口的差距有多大。
大模型 精选 · 原创 · 10-9 阅读 245·访客 240
Transformer 架构全景：从 2017 年的原点到各大厂变种
从 Attention Is All You Need 的 encoder-decoder 原型，到 MLA、MoE、iRoPE 缀满一身的 2026 旗舰，本文系统拆解基础 Transformer 的架构原理，梳理八年来 KV 压缩、位置编码、稀疏专家三条演进主线，并给出 DeepSeek、Llama、Qwen、Gemini 等旗舰架构的横向对比地图。
大模型 精选 · 原创 · 10-9 阅读 676·访客 604
长上下文的注意力之战：滑窗、外推与原生稀疏的三线会战
2026 年，1M token 上下文已成为头部旗舰的标配，但 RULER 与 NoLiMa 评测反复证明「宣称上下文」远大于「有效上下文」。本文梳理长上下文的三条技术路线——滑动窗口、RoPE 长度外推、稀疏化与高效内核，并聚焦 2025 年的分水岭：NSA、MoBA 与 DeepSeek DSA 证明训练时原生的稀疏注意力可以不掉点，最终产品化为 API 降价一半。
研究前沿 精选 · 用户投稿 · 10-9 阅读 496·访客 494
A single prompt was enough to hijack every AI agent in an AWS account, Zenity researchers found
Oct 8, 2026 Nano Banana Pro prompted by THE DECODER Key Points A chain of vulnerabilities in Amazon's Bedrock AgentCore …
智能体 The Decoder · 10-8 阅读 64·访客 64
The Download: AI roadblocks for humanoids and portable rubber dams
This is today's edition of* *The Download*,*our weekday newsletter that provides a daily dose of what's going on in the …
智能体 MIT Technology Review · 10-8 阅读 56·访客 55
AI-powered hacking tools enabled a likely single attacker to breach multiple South Korean banks
Oct 8, 2026 Crowdstrike reports on an infrastructure hack in South Korea.** A suspected Chinese-speaking attacker hit mu…
行业动态 The Decoder · 10-8 阅读 26·访客 26
Anthropic Releases Claude Haiku 5.5: A Small Model With 1M Context Priced at $0.10 per Million Input Tokens
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
大模型 MarkTechPost · 10-8 阅读 40·访客 40
Perplexity AI Releases pplx-embed-v2-late: A 0.6B Edge Model and a 9B Model Scoring 92.4% on MADQA
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
行业动态 MarkTechPost · 10-8 阅读 28·访客 28