You need to scale your work up and out

最新推荐文章于 2025-12-18 17:30:06 发布

原创最新推荐文章于 2025-12-18 17:30:06 发布 · 771 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#工作 #竞争 #软件 #技术 #服务器

读书笔记专栏收录该内容

273 篇文章

订阅专栏

本文介绍了两个概念：Scaleup 和 Scaleout，分别指提升个人技术的独特性和将成果规模化的能力。要成为不可替代的人才并实现个人价值的最大化，需关注这两方面的发展。

借用服务器后端开发的2个术语。

Scale up：这意味着必须提高你的工作的技术难度——不是别的什么人随随便便也能完成你做的工作。这让你的核心竞争力不可替代。

Scale out：你的工作最好是能够通过无成本的复制卖给大量顾客。在一个大企业里当螺丝钉显然不行。写一个自己的软件或者写一本书也许可能。

确定要放弃本次机会？

福利倒计时

: :

立减 ¥

普通VIP年卡可用

立即使用

志_祥

关注关注

0
点赞
踩
0

收藏

觉得还不错? 一键收藏
0
评论
分享

复制链接

分享到 QQ

分享到新浪微博

扫一扫
举报

举报

专栏目录

多智能体强化学习：Learning to Cooperate, Compete, and Communicate

weixin_41697507的博客

09-28

1572

Multiagent environments where agents compete for resources are stepping stones on the path to AGI. Multiagent environments have two useful properties: first, there is a natural curriculum—the difficulty of the environment is determined by the skill of your

Attention Is All You Need

热门推荐

程永强

05-11

4万+

Attention Is All You Need 注意力机制是你需要的全部 Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, Illia Polosukhin attention [ə'tenʃ(ə)n]：n. 注意，关注，注意力，关心 int. 注意，立正 attention mechanism：注意力机制 Computer Science，CS：

参与评论您还未登录，请先登录后发表或查看评论

So, you want to work in security？ By Parisa Tabriz

汤姆lazy的优快云博客

08-26

908

Every once in a while, I’ll get an email from an eager stranger asking for advice on how to have a career in security (computer, information, cyber… whatever). This is great! We need more passionate, creative, hard-working people that want to work on maki.

论文解读：Attention Is All You Need

weixin_42046845的博客

12-12

6431

谷歌于2017年发布论文《Attention Is All YouNeed》，提出了一个只基于attention的结构来处理序列模型相关的问题，比如机器翻译。相比传统的CNN与RNN来作为encoder-decoder的模型，谷歌这个模型摒弃了固有的方式，并没有使用任何的CNN或者RNN的结构，该模型可以高度并行的工作，相比以前串行并且无法叠加多层、效率低的问题。那么Transorformer可以高度并行的工作，所以在提升翻译性能的同时训练速度也特别快。

论文精读--Attention is all you need

m0_73202283的博客

01-23

2691

主要的序列转导模型是基于复杂的循环或卷积神经网络，包括一个编码器和一个解码器。表现最好的模型还是通过注意机制连接编码器和解码器。我们提出了一个新的简单的网络架构，transformer，完全基于注意力机制，完全摒弃循环和卷积。在两个机器翻译任务上的实验表明，这些模型在质量上更优越，同时更具并行性，并且需要更少的训练时间。我们的模型在WMT 2014英语-德语翻译任务上实现了28.4 BLEU，比现有的最佳结果(包括集合)提高了2个BLEU以上。

paper:Attention Is All You Need（源码篇）

AI蜗牛之家

01-12

2665

继上篇paper:Attention Is All You Need之后，我们针对official提供的源码进行解析。本次源码分析针对的是tensorflow/models，同时还有一份带有详细注释的源码，在小博的github里面。(好久没整理这么详细的帖子了。。。) 本博主题思路参考：Attention Is All You Need(注意力模型），并在此基础上做了修改，并增加了相关源码解析~...

Attention Is All You Need-论文解读（不含实验）

qq_55453054的博客

12-30

1908

在这些模型中，关联来自两个任意输入或输出位置的信号所需的操作数量随着位置之间的距离而增加，对ConvS2S是线性的，对于ByteNet是对数的。这种掩蔽，再加上输出嵌入偏移一个位置的事实，确保了对位置i的预测只能取决于小于的位置处的已知输出。解读:在卷积神经网络中，卷积在面对长序列时，会产生累积的效果，当累计到一定程度才能到达序列的结尾，此外，卷积还具有多通道输出的效果。解读:在下面的解码器中，我们要注意红色框中的内容，它实现了在t时间的时候不会看到t时间之后的输入，以保证训练和预测的时候行为一致。

Scaling Up And Out

Mac Track

01-06

733

Most attention today is focused on adding nodes or cloud instances to scale out systems. Guest editor Nikita Shamgunov emphasizes the importance of scaling systems vertically as well. Recentl

You and Your Research

BALANCE

09-02

1479

Richard Hamming ``You and Your Research'' Transcription of the Bell Communications Research Colloquium Seminar 7 March 1986

You AND Your Research

senduo的专栏

03-20

2726

Talk at Bellcore, 7 March 1986 The title of my talk is, ``You and Your Research.'' It is not about managing research, it is about how you individually do your research. I could give a talk on the o

suse ha for sap scale-up性能优化场景安装配置

jetliu05的博客

07-05

1921

suse ha for SAP安装配置

What you can try is that you can divert the numerical explicitness of your features. Let me explain:\nSimilar to my answer in the previous topic; the regression algorithm will use the value from the time-window you give as a sample, to minimize the error. Let’s assume you are trying to predict the closing price of BTC at time t. One of your features consists of previous closing prices and you are giving a time-series window of last 20 inputs from t-20 to t-1. A regressor probably will learn to choose the closing value at time step t-1 or t-2 or a close value in this case, cheating. Think like that: if closing price was $6340 at t-1, predicting $6340 or something close at t+1 would minimize the error at strongest. But actually the algorithm did not learn any patterns; it just replicates, so it basically does nothing but accomplishing its optimization duty.\nThink analogously from my example: By diverting the explicitness, what I mean is that: do not give the closing prices directly, but scale them or do not use explicit ones at all. Do not use any features explicitly showing the closing prices to the algorithm, do not use open, high, low etc for every time step. You will need to be creative here, engineer the features to get rid of explicit ones; you can give squared close differences (regressor can still steal from past with linear differences, with experience), its ratio to volume. Or, can make the features categorical by digitizing them in a manner that would make sense to use. The point is do not give direct intuition to what it should predict, only provide patterns for algorithm to work on.\nA faster approach may be suggested depending on your task. You can do multi-class classification if predicting how much percent of change that your labels is enough for you, just be careful about class imbalance situations. If even just the up/down fluctuations are enough for you, you can directly go for the binary classification. Replication or shifting problems are only seen at the regression tasks, if you are not leaking data from training to the test set. If possible, get rid out of regression for time-series windowed applications.

最新发布

12-19

This smooths out the price data and can help in identifying trends. ```python import pandas as pd # Assume df is a DataFrame with a 'Close' column representing BTC closing prices df = pd.DataFrame...

Edit UVs & crop textures. 可编程渲染管线（SRP）兼容性 Unity可编程渲染管线（SRP）是让您能够通过C#脚本控制渲染的一个功能。SRP是支撑通用渲染管线（URP）和高清渲染管线（HDRP）的底层技术。其他兼容性信息 Editor only. This is NOT for runtime use! 描述 Links: Manual Move, Rotate and Scale UVs with ease. And while you are at it you can also move your texture pixels along. Finally crop the texture and you are done. !! ⚠️ IMPORTANT ⚠️ !! Version 2 pack a lot of significant updates. Please delete to old version of the tool BEFORE installing this new version. 😎 Easy to use No need to do the cumbersone selection of many vertices. Simply paint on your mesh and all the painted triangles are selected. 💫 UV Editing Move, Rotate, Scale your UV howevery you like. The non-destructive workflow ensures all your original assets (meshes, materials, textures) are left untouched. The tool will generate and assign new ones automatically under Assets/UVEditorAssets/. 😺 Texture Editing The texture editing mode allows you to move pixels within your texture based on the selected triangles. Very handy to clean up messy layouts. And once you are done you can use the crop feature or trim the fat. Easy. 📚 Documentation Check out https://kamgam.com/unity/UVEditorManual.pdf ✔️ Full Source Code I publish this asset with full source code because as a developer I know how annoying it is to not have source code access. If you like it then please remind others to purchase the plugin. Don't just copy and paste it everywhere. Thank you ❤️ ✔️ Supports Unity 2021.3+, 2022, 2023, Unity 6, ... LTS It may work with earlier versions too but those are not tested anymore. 👇 Read this before you buy 👇 ⚠️ This is an EDITOR ONLY tool. It does not have runtime support. ⚠️ This is NOT a UV mapper tool. The tool can be used to EDIT existing UV layouts (see video) BUT if your model does not have UVs yet and needs a proper layout then this most likely is not the tool for it. ⚠️ Unity 2021.3 or higher is required (LTS releases strongly recommended). ⚠️ It may be slow on meshes with > 60.000 vertices. If you run into performance issues then please set the UV Filter to "Selected" to limit the UV display to the selection only. ✍️ Usage 1) To to Tools/UV Editor (or use the tools side bar) 2) Open the UV Editing Window 3) Select your Object and UVs 4) Move, scale or rotate them. 5) Hit "Apply" at the bottom of the UV Editor window. 6) Done. 😎 HINT 1: Editing single vertices is possible. Just select on of the adjacent triangles, switch to vertex mode and then delselect the vertices you do not want to move. 🧐 HINT 2: In crop mode notice that only the UVs that use the same texture are cropped. 🤠 HINT 3: Undo / Redo works only if the UV Editor window is focused (blue highlight at the top). ☎️ Support If you have any questions please write to office[at]kamgam.com. Please be patient, replies may take a few days. Please include your Asset Store Order Nr and the exact Unity version in your request. Please also try upgrading to the most recent LTS version before asking. Thank you. ❤️ If you like this asset then maybe you'll like these too: ⚙ Settings Generator One Settings UI + API for all render pipelines. ✂ Mesh Extractor 2 Separate your assets into useful parts within seconds. 🚧 Exclude from build Exclude any files and directories from build. 🏹 Power Pivot Edit, Rotate, Scale, Snap & Pivot Editor 我买了这个插件但是我不知道他什么意思怎么使用

04-03

1. **UV编辑三件套**：移动(Move)、旋转(Rotate)、缩放(Scale)现有UV布局 2. **纹理像素调整**：通过选中UV区域直接移动纹理像素 3. **智能裁剪**：自动裁剪掉UV范围外的空白纹理区域 4. **非破坏性编辑**：自动生成...

Basic Knowledge You Need to Have for Golang（Golang 基础知识概括）

人生如路

01-17

1980

Basic Knowledge You Need to Have forGolang This article covers the knowledge which is important to know as a golang developer The main reference is a tour of go, build web application with golang a...

软考架构师实战：Spring Boot 3.5 + DeepSeek 开发 AI 应用，上线 24 小时数据复盘（2C1G 服务器抗压实录）

java_yuan的专栏

12-13

413

兄弟们，我又来了。软件架构师实战，我那台服务器是香港的2c1G的，技术栈是 Spring Boot 3.5 + HTMX + DeepSeek 的技术栈。通过首日流量情况、服务器资源占用（CPU/内存）以及遇到的性能瓶颈与优化方案。

服务器被DDOS跟CC攻击了怎么处理,如何抵御攻击?

YOKEhn的博客

12-18

228

而 CC 攻击更偏向“应用层”，它模拟真实用户不断访问动态页面、接口或登录行为，消耗服务器计算资源，往往在带宽并未完全打满的情况下就能造成服务不可用。真正成熟的防御体系，并不是完全杜绝攻击，而是在攻击发生时，业务依然能够保持可控、可恢复、可持续。CDN 节点分布在各地，天然具备分散流量的能力，攻击流量会被分摊到多个节点，减少单一服务器的压力。在应急阶段结束后，更重要的是建立长期、系统化的防御策略。无论是企业官网、电商平台，还是游戏、接口服务，一旦被攻击，轻则访问缓慢、用户流失，重则业务中断、数据风险增加。

预算49800，99800，299800能买到什么样的算力服务器

yilinwangluo的博客

12-18

515

预算越高，GPU的显存、算力（浮点运算能力）和网络互联性能越强，适用的任务规模（模型大小、数据量）也越大。49,800元是“能用”的入门门槛，99,800元是“好用”的企业级标准，299,800元则是“顶尖”的大模型/超算级配置。

超微服务器的ipmi调整风扇运行模式

sqlora的专栏

12-15

524

root@node15:/disk2/Qwen2.5-72B-Instruct-GPTQ-Int4# ipmitool lan print Set in Progress : Set Complete Auth Type Support : NONE MD2 MD5 PASSWORD Auth Type Enable : Callback : MD2 MD5 PASSWORD : User : MD2 MD

Google推出托管MCP服务器，让AI Agent轻松接入其工具生态

2501_94005722的博客

12-18

289

随着AI Agent生态系统的快速发展，Google的这一举措为开发者提供了更简单、更安全的方式来连接AI Agent与Google Cloud服务。通过标准化的MCP协议和完全托管的服务器，Google正在降低AI Agent集成的门槛，同时确保企业级的安全和治理要求得到满足。AI Agent被宣传为能够规划旅行、回答商业问题、解决各种问题的解决方案，但让它们与聊天界面之外的工具和数据协同工作一直是个难题。开发者需要拼凑各种连接器并保持运行，这种方法既脆弱又难以扩展，还会带来治理上的麻烦。

You will need to adjust your conda configuration to proceed.

05-31

This error message suggests that there is an issue with your conda configuration that is preventing the program from running properly. To proceed, you may need to adjust your conda configuration. One possible solution is to try running the following command in your terminal: ``` conda init ``` This command initializes your conda shell configuration, which may resolve the issue you are encountering. If this does not resolve the issue, you may need to check your conda configuration settings and adjust them as needed. You can find more information on how to do this on the Conda documentation website or by searching for specific solutions to the error message you are seeing. Let me know if you have any further questions or concerns!