Artifical Intelligence Blog

Wake instructors utilizing AI in the class set tone for safe, clever knowing

In the heart of Wake County, teachers are setting the stage for a revolution in education. By integrating Artificial Intelligence (AI) into their classrooms, they’re creating a safer, smarter learning environment. This innovative approach is transforming the way students learn, making education more personalized, engaging, and effective. Stay tuned as we delve into this exciting development in the world of AI.

Malaysia Announces Shariah Compliant Artificial Intelligence Framework

Malaysia is making strides in the AI world with its recent announcement of a Shariah Compliant Artificial Intelligence Framework. This unique approach integrates AI technology with Islamic principles, opening new possibilities for AI applications in sectors such as finance, healthcare, and education. Stay tuned as we delve into this groundbreaking development.

一文了解Transformer全貌(图解Transformer)

打个小广告 ☻,知乎专栏《大模型前沿应用》的内容已经收录在新书《揭秘大模型:从原理到实战》中。感兴趣的朋友可以购买,多谢支持!♥♥ 自2017年Google推出Transformer以来,基于其架构的语言模型便如雨后春笋般涌现,其中Bert、T5等备受瞩目,而近期风靡全球的大模型ChatGPT和LLaMa更是大放异彩。网络上关于Transformer的解析文章非常大,但本文将力求用浅显易懂的语言,为大家深入解析Transformer的技术内核。 前言 Transformer是谷歌在2017年的论文《Attention Is All You Need》中提出的,用于NLP的各项任务,现在是谷歌云TPU推荐的参考模型。网上有关Transformer原理的介绍很多,在本文中我们将尽量模型简化,让普通读者也能轻松理解。 在机器翻译中,Transformer可以将一种语言翻译成另一种语言,如果把Transformer看成一个黑盒,那么其结构如下图所示: 将法语翻译成英语 那么拆开这个黑盒,那么可以看到Transformer由若干个编码器和解码器组成,如下图所示: 继续将Encoder和Decoder拆开,可以看到完整的结构,如下图所示: Transformer整体结构(引自谷歌论文) 可以看到Encoder包含一个Muti-Head Attention模块,是由多个Self-Attention组成,而Decoder包含两个Muti-Head Attention。Muti-Head Attention上方还包括一个 Add & Norm 层,Add 表示残差连接 (Residual Connection) 用于防止网络退化,Norm 表示 Layer Normalization,用于对每一层的激活值进行归一化。 假设我们的输入包含两个单词,我们看一下Transformer的整体结构: Transformer整体结构(输入两个单词的例子) 为了能够对Transformer的流程有个大致的了解,我们举一个简单的例子,还是以之前的为例,将法语”Je suis etudiant”翻译成英文。 Transformer输入表示 输入X经过Encoder输出编码矩阵C Transformer Decoder预测 上图Decoder接收了Encoder的编码矩阵,然后首先输入一个开始符 “”,预测第一个单词,输出为”I”;然后输入翻译开始符 “” 和单词 “I”,预测第二个单词,输出为”am”,以此类推。这是Transformer的大致流程,接下来介绍里面各个部分的细节。 2. Transformer的输入表示 Transformer中单词的输入表示由单词Embedding和位置Embedding(Positional Encoding)相加得到。 Transformer输入表示 2.1 单词Embedding 单词的Embedding可以通过Word2vec等模型预训练得到,可以在Transformer中加入Embedding层。 2.2 位置Embedding Transformer 中除了单词的Embedding,还需要使用位置Embedding

World Monetary institution Country and Lending Groups – World Monetary institution Data Back Desk

For the current 2026 fiscal year, low-income economies are defined as those with a GNI per capita, calculated using the World Bank Atlas method, of $1,135 or less in 2024; lower middle-income economies are those with a GNI per capita between $1,136 and $4,495; upper middle-income economies are those with a GNI per capita between

This AI Startup Is Training The Next Generation Of Accountants

In a world where AI is reshaping industries, a groundbreaking startup is harnessing its power to train the accountants of tomorrow. This innovative venture is revolutionizing the financial sector, making complex calculations and financial forecasting as easy as a click of a button. Stay tuned as we delve into how this startup is making waves in the world of finance.

逆天|详细说说苹果M4、M4 Pro和M4 Max

上周,苹果发布了搭载最新M4芯片的三款Mac。 M4是苹果自己设计的芯片,ARM架构,采用台积电第二代3nm制程制造。 其实它早在24年5月就在iPad Pro上搭载了,只是最近才把它放在了Mac上。 这一代同上一代M3一样,是3nm制程。但M4的3nm是台积电的第二代3nm工艺,在能效表现上又跨了一大步,并且竟然没扯到蛋。 所以,这一代Mac的性能相比上一代提升了30%,MacBook Pro续航也大大增加,达到了24小时,让俺们牛马在实现724工作制的道路上又近了一步。 M4一共有三个版本,M4、M4 Pro以及M4 Max。 即所谓的中杯、大杯、超大杯。 满血版的普通M4有10个CPU核心,10个GPU核心和16个NPU核心。 其中,10个CPU核心包含4颗性能核心和6颗能效核心。 这颗芯片搭载在基础款的MacBook Pro、Mac mini和iMac上。 不过要注意,iMac最低配上的M4是残血版,它的CPU和GPU都只有8核。 估计未来的M4 MacBook Air低配款同样也是残血M4。 这是在geekbench上,满血版10核M4的跑分: 因为工艺进步,M4的单核跑分大幅领先M3的3100,史上最高。 从多核跑分上来看也是不得了,10核的M4达到了12核M2 Max的水平。 这里简单说下单核跑分和多核跑分的意义。 单核跑分反映其中的一个CPU核心的性能。单核分数越高,在处理单线程任务的时候,性能就越好,用起来越丝滑。 日常的很多工作大多属于是这种类型的,比如上上网、做做表什么的。 因此,普通款的M4对于轻度用户来说是个很好的选择。如果不搞编程、不剪视频啥的,我觉得还可以等等明年更便宜的MacBook Air。 而多核跑分反映的就是所有CPU集火输出的能力,对于专业用户来说,比如剪视频、编译代码啥的,多核跑分就很重要。 因此,单论CPU性能,M4用起来绝对比M2 Max爽。 二、大杯:M4 Pro 现在我们来看看中档的M4 Pro。 M4 Pro其实有两款,差价1500块: 12核CPU、16核GPU 14核CPU、20核GPU 满血版的是14核CPU和20核GPU。CPU核心多了4个,但GPU数量相比普通M4翻倍了。 M3 Pro的内存带宽很高,达到了273GB/s,相比M3 Pro提升了75%。 高带宽意味着短时间内可以向内存读写更多的内容,在处理重度任务时比较有优势,比如视频渲染、大模型训练等等。 另外,M4 Pro支持雷雳5接口,而普通M4只支持雷雳4。 「雷雳」就是英特尔的雷电技术,用的是USB Type-C接口。雷雳5传输速率可以达到120 Gbit/s,非常快,是雷雳4的三倍。 如果你有雷雳5设备,传输数据的时候就会非常爽。 但雷雳5的设备价格非常贵,就连一根数据线都可能要上百块,你的钱包可能会非常不爽。 我们来看看M4 Pro的跑分: 单核就不说了,所有M4都一样。

Man made Intelligence

Authors What is AI?  Artificial intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy. Applications and devices equipped with AI can see and identify objects. They can understand and respond to human language. They can learn from new information and experience. They can

Inteligencia artificial

Imagen generada por el modelo de IA de generación de imágenes DALL•E. Vídeo explicativo de 6:47 min, en idioma euskera (con subtítulos en castellano) sobre la inteligencia artificial, incluyendo secciones sobre los dilemas éticos. La inteligencia artificial (abreviado: IA), en el contexto de las ciencias de la computación, es una disciplina y un conjunto de capacidades

Kathmandu Questions and Answers Class 9 English Beehive Chapter 10 NCERT Alternate choices

NCERT Solutions for Class 9 English Class 9 English Beehive Chapter 10 Kathmandu Question Answer English Class 9th Beehive Chapter 10 Kathmandu Question Answers Kathmandu Thinking About the Text Question 1. On the following map mark out the route, which the author thought of but did not take, to Delhi. Answer: The route the author

No topic its spectacular output, generative AI doesn’t like a coherent working out of the field

Large language models can do impressive things, like write poetry or generate viable computer programs, even though these models are trained to predict words that come next in a piece of text. Such surprising capabilities can make it seem like the models are implicitly learning some general truths about the world. But that isn’t necessarily

Check out the Newest Amazon gadgets

Master AI & ML Today!

Python Machine Learning: Machine Learning ​and Deep Learning with Python, scikit-learn, and TensorFlow 2, 3rd Edition Python Machine Learning, Third Edition is a comprehensive guide that melds theory‍ with⁤ practical applications in machine learning and deep learning...

read more

Check out our other Contents in Specific areas of AI

Artificial Intelligence Ethics, Trustworthiness and Legislation

Artificial Intelligence Model News

Artificial Intelligence Prompt Engineering

AI Regulatory Compliance

Data And Document Processing

AIM-E can futureproof my companyAI Data and Document ProcessingAIM-E's Data and Document Processing solutions harness the power of AI to unlock valuable insights from unstructured data. By automating the extraction, classification, and analysis of information from...

Cloud Consulting

Cloud ConsultingRecognizing the transformative potential of cloud computing, AIM-E offers comprehensive cloud consulting services tailored to each business's unique needs. From migrating workloads to optimizing cloud infrastructure and integrating multiple cloud...

Managing AI With AI using APIs

Discover how advanced AI systems can monitor, control, and enhance other AI systems to ensure compliance, accuracy, and safety in enterprise environments.Artificial Intelligence Made Easy is led by technology visionary Jim Barnebee, a top UX in AI executive recognized...

Basic Model Checks

Access and utilize powerful APIs to check and manage AI model data efficiently.Discover how our tools can enhance your AI interactions effortlessly.Our platform provides a comprehensive suite of APIs designed to facilitate seamless interaction with AI model data. By...

Our CEO wrote a children’s book using AI- Read for Free here on Kindle unlimited

Azure ai throwing carrier ask failed error 400

Azure ai throwing carrier ask failed error 400

I upload same file on the azure portal and try it there seems returning correct result. Portal provides search results as “context” in the prompt. So, Here AI model is primed with relevant data to answer the question accurately. When you pass the search results manually, these preprocessing and fine-tuning steps are likely missing, leading

read more
“An AI future that honors dignity for everybody”

“An AI future that honors dignity for everybody”

Ben Vinson III, president of Howard University, made a compelling call for artificial intelligence to be “developed with wisdom,” as he delivered MIT’s annual Karl Taylor Compton Lecture on campus Monday.  The broad-ranging talk posed a series of searching questions about our human ideals and practices, and was anchored in the view that, as Vinson

read more
The employ of generative AI, researchers develop compounds that might perchance extinguish drug-resistant bacteria

The employ of generative AI, researchers develop compounds that might perchance extinguish drug-resistant bacteria

With help from artificial intelligence, MIT researchers have designed novel antibiotics that can combat two hard-to-treat infections: drug-resistant Neisseria gonorrhoeae and multi-drug-resistant Staphylococcus aureus (MRSA). Using generative AI algorithms, the research team designed more than 36 million possible compounds and computationally screened them for antimicrobial properties. The top candidates they discovered are structurally distinct from any

read more

Trí tuệ nhân tạo

Trí tuệ nhân tạo (TTNT) (tiếng Anh: Artificial intelligence , viết tắt: AI ) là khả năng của các hệ thống máy tính thực hiện các nhiệm vụ liên quan đến trí thông minh của con người, như học tập, suy luận, giải quyết vấn đề, nhận thức và đưa ra quyết định. Đây là

read more
2025年 8月 CPU天梯图(更新锐龙9 9950X3D)

2025年 8月 CPU天梯图(更新锐龙9 9950X3D)

CPU天梯图更注重综合性能,只具有参考意义,选择CPU要根据需求选择。 生产力用户更偏重多核性能,可以参考综合性能天梯图 游戏用户更偏重单核性能,综合性能天梯图参考意义不大 从左到右依次是1080P/2K/4K分辨率下,CPU游戏性能排名,测试显卡均为4090,目前最强游戏U为9800X3D。 左1080P右4K分辨率下,CPU游戏平均帧数排名,测试显卡均为5090所以目前数据较少,更新最新上线的9950X3D。 数据来源:TechpowerUP 桌面端CPU综合性能天梯图: 作者:秋刀鱼半藏(原帖地址) 生产力性能排名 (R9 9950X3D更新了测试平台,目前参考数据较少) 机器学习/AI:Stable Diffusion、图像分类、图像放大 (GPT2、图像放大、图像分类) 编解码性能:AV1、H.264、H.265 渲染性能:Cinbench 2024、Blender、V-Ray CPU购买推荐: 在非极端情况下,CPU主板一起买价格都是要低于单独买的,而且CPU本身极不容易损坏,所以一般都推荐在淘宝/拼多多入手,618/双11京东优惠力度大的时候优先京东自营入手。 ①办公、家用核显U:5600G 5600G 6核显12线程,核显性能也还可以,玩一些网游,应对家用办公场景都没问题,主板搭配上推荐B450或者A520,这里推荐的是5600G+微星A450-A PRO。 ②游戏性价比:12400F(同12490F)、5600 12400F(12490F) 6核显12线程,游戏性价比神U,历经三代不衰,性价比很高,4~6K价位主机最热门的选择。 提起12400F不能不提5600,5600也是性价比极高的存在,这两个可以互换 如果这里预算略微高一些,可以CPU升级成13400F/13490F,12600KF目前价格涨太多,不值得入手了。 ③游戏进阶:7500F 7500F的游戏性能是很强的,基本可以和13600KF干,不过多核性能就差一些了,如果单纯是为了玩游戏,特别是网游和1080P分辨率,7500F是很值得入手的,这里推荐搭配华硕A620M AYW主板。 ④游戏+生产力性价比:14600KF 14600KF目前的价格性价比极高,很挺值得入手的,特别是对CPU多核性能有一定要求的朋友。 ⑤最强游戏U:9800X3D 目前最强的两款游戏CPU。 不过两个目前溢价都较为严重,1080P高帧显示器用户和预算拉满的用户推荐入手,否则不建议入手。 ⑥高端生产力CPU推荐: U7 265K: 目前生产力性价比神U,生产力用户无脑可冲。 U9 285K: 目前生产力天花板,但是性价比不如U7。 9950X&9950X3D: 生产力性能不错,X3D系列缓存大在部分应用有优势。 显卡天梯图 2025年 7月 显卡天梯图(更新RTX 5060) 电脑配置推荐: Gyusang:2025年装机电脑配置推荐(配置单可以直接照抄) 相关阅读: CPU: CPU选购指南及推荐

read more

Join The Newsletter

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy policy and terms and conditions on this site