NEWS / 资讯
资讯
只聚合标题、AI 摘要与原文链接。列表按发布时间倒序;同源同主题的重复报道会被折叠成簇。
-
超频三融核系列 27 英寸 AI 模组化整机开售:酷睿 Ultra 5 125U 处理器,首发 11999 元
来源摘要 IT之家 9 月 22 日消息,超频三融核系列 27 英寸 AI 模组化整机现已开售,新品采用酷睿 Ultra 5 125U 处理器,专为本地轻量 AI 部署设计,首发售价 11999 元。
-
安卓手机也能用 Pr 剪片了,Adobe Premiere 移动版登陆谷歌 Play 商店
来源摘要 IT之家 9 月 22 日消息,iOS 版上线一年多后,Adobe Premiere 移动应用今天(22 日)晚间正式登陆安卓平台。与 iOS 版相同,安卓版 App 同样可以免费下载,生成式 AI 功能则需通过内购使用。分割、修剪、裁剪、合并视频,调整播放速度,以及添加文字、特效和旁白等音频层等剪辑功能全部免费。用户无需付费即可不限次数导出无水印 4K 视频,并能…
-
Show HN: JevBench, a reproducible benchmark for typed decision models
来源摘要 Hi HN! I built JevBench because Jev kicks ass, and the world deserves to know how the serious open source and fake lookalike projects really perform in comparison.Jev-class models return bounded choices and probabilities instead of text, and are disruptively faster and cheaper th…
-
Show HN: Training a model to identify AI web content from structure alone
来源摘要 Hey HN! We’re Vincent and Jochen from Sitefire (https://sitefire.ai). We have been working together for years, with backgrounds in RL/optimization at Stanford and software engineering from Technical University Munich (TUM).With Sitefire (YC W26), we help marketing teams get recom…
-
OPPO 推出“AI 心力球”可穿戴新品,体验价 499 元
来源摘要 IT之家 9 月 22 日消息,OPPO 今天召开新品发布会,除了推出 Find X10 系列手机新品以外,OPPO 本次还带来了全新品类穿戴产品“AI 心力球”。新品体验价为 499 元,主打卖点是“AI 成长教练”,能够帮助用户思考,把事情做成。即日起开启用户招募。IT之家在发布会现场了解到,OPPO AI 心力球是一款别在衣服领口的穿戴产品…
-
“十五五”时期,我国将推动北斗产业规模突破 1 万亿元
来源摘要 IT之家 9 月 22 日消息,央视新闻从国家发展改革委今天(22 日)举行的新闻发布会上了解到,“十五五”时期,我国将推动北斗产业规模五年内突破 1 万亿元,实现北斗规模应用市场化、产业化、国际化新突破。国家发展改革委新闻发言人李超介绍,“十五五”时期,我国将推进高端芯片、关键零部件、融合算法等关键核心技术攻关,加强北斗与人工智能、卫星互联网等技术融合创新,推动卫星导航基础设施布局更加合理,培育壮大优质企业,筑牢安全可靠的时空底座。同时,加快推动《中华人民共和国卫星导航条例》出台实施,持续推动技术交流、人才培育、联合示范等务实合作项目落地,积极参与卫星…
-
路透社民调:近四分之三美国人认为 AI 公司在防止灾难方面做得不够
来源摘要 IT之家 9 月 22 日消息,路透社 / 益普索(Ipsos)最新民调显示,73% 的美国人担心,AI 公司为防止人工智能严重危害社会所采取的措施仍然不够。IT之家从今天(22 日)晚间公布的数据获悉,39% 的受访者认为 AI 正在给社会带来负面影响,高于上月的 36%,创下自 3 月开始调查这一问题以来的最高水平。只有 11% 认为 AI 带来了积极影响,其余受访者表示不确定或没有作答。许多美国人也不相信 AI 企业有能力管好自己。多数受访者认为,联邦政府应该在制定 AI 安全标准方面发挥重要作用。
-
联合国 AI 专家反对 AI 末日论,呼吁理性讨论
来源摘要 IT之家 9 月 22 日消息,联合国下属人工智能委员会的多名专家于当地时间周一发出提醒,反对业内部分从业者针对 AI 风险所发表的世界末日式言论。独立机构国际 AI 科学专家组(International Scientific Panel on AI)成员、谷歌 DeepMind 高管若埃尔 · 巴拉尔(Joelle Barral)表示:“身为科研工作者,我们应当投入精力开展科学研究,厘清哪些…
-
美银等多家银行警告:“AI 代购”或增加诈骗、欺诈和数据隐私泄露风险
来源摘要 IT之家 9 月 22 日消息,据路透社今天(22 日)傍晚报道,国民西敏银行、美国银行等多家银行警告,让 AI 智能体代替消费者进行网上购物,可能增加诈骗、欺诈和数据隐私泄露的风险。OpenAI、Anthropic、谷歌和 Meta 等科技企业正大力把 AI 聊天机器人推向购物场景,希望未来消费者可以让 AI 智能体挑选商品并直接代为下单,零售商则开始争相影响聊天机器人的商品推荐。
-
MOPD: Multi-Teacher On-Policy Distillation for Capability Integration in LLM Post-Training
来源摘要 Modern large language models (LLMs) rely on reinforcement learning during post-training to push specific capabilities, yet integrating multiple capabilities into one model remains hard. Existing methods, such as Off-Policy Finetune and Mix-RL, are either inefficient or lose perfo…
-
MiMo-V2-Flash Technical Report
来源摘要 We present MiMo-V2-Flash, a Mixture-of-Experts (MoE) model with 309B total parameters and 15B active parameters, designed for fast, strong reasoning and agentic capabilities. MiMo-V2-Flash adopts a hybrid attention architecture that interleaves Sliding Window Attention (SWA) with…
-
MiMo-Audio: Audio Language Models are Few-Shot Learners
来源摘要 Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions. GPT-3 has shown that scaling next-token prediction pretr…
-
ARL-Tangram: Unleash the Resource Efficiency in Agentic Reinforcement Learning
来源摘要 Agentic reinforcement learning (RL) has emerged as a transformative workload in cloud clusters, enabling large language models (LLMs) to solve complex problems through interactions with real world. However, unlike traditional RL, agentic RL demands substantial external cloud reso…
-
Your First API Call | DeepSeek API Docs
来源摘要 The DeepSeek API uses an API format compatible with OpenAI/Anthropic. By modifying the configuration, you can use the OpenAI/Anthropic SDK or softwares compatible with the OpenAI/Anthropic API to access the DeepSeek API.
-
🚀 DeepSeek V2.5: The Grand Finale 🎉 | DeepSeek API Docs
来源摘要 🌐 Internet Search is now live on the web! Visit https://chat.deepseek.com/ and toggle “Internet Search” for real-time answers. 🕒
-
🚀 DeepSeek-R1-Lite-Preview is now live: unleashing supercharged reasoning power! | DeepSeek API Docs
来源摘要 🔍 o1-preview-level performance on AIME & MATH benchmarks.
-
DeepSeek-V2.5: A New Open-Source Model Combining General and Coding Capabilities | DeepSeek API Docs
来源摘要 We’ve officially launched DeepSeek-V2.5 – a powerful combination of DeepSeek-V2-0628 and DeepSeek-Coder-V2-0724! This new version not only retains the general conversational capabilities of the Chat model and the robust code processing power of the Coder model but also better ali…
-
DeepSeek API introduces Context Caching on Disk, cutting prices by an order of magnitude | DeepSeek API Docs
来源摘要 In large language model API usage, a significant portion of user inputs tends to be repetitive. For instance, user prompts often include repeated references, and in multi-turn conversations, previous content is frequently re-entered.
-
DeepSeek API Upgrade | DeepSeek API Docs
来源摘要 Now Supporting Chat Prefix Completion, FIM, Function Calling and JSON Output
-
AI 助力古籍修复:Apollo 大语言模型问世,专门用于解读古希腊莎草纸残片
来源摘要 IT之家 9 月 22 日消息,据 《连线》(WIRED)报道,全球各地高校图书馆保存着数十万份古希腊莎草纸残片,不少残片损毁严重,其内容或许已无法解读;对于其余尚存修复希望的文本,学者只能依靠人工逐条补全缺失的词句完成复原。为加快这项繁重的工作,研究人员开始借助人工智能。奥地利科学院将于当地时间周三正式发布了全球首款面向古希腊语的高级大语言模型。该项目由奥地利科学院联合法国 AI 实验室 Mi…