It's kind of like ChatGPT, except it's not.

它有點像 ChatGPT，只不過不是。

Everything about it is running right here on my computer.

它的一切都在我的電腦上運行。

I'm not even connected to the internet.

我甚至都沒有聯網。

This is private, contained, and my data isn't being shared with some random company.

這是私人的、保密的，我的數據不會被隨便一家公司共享。

So in this video, I want to do two things.

是以，在這段視頻中，我想做兩件事。

First, I want to show you how to set this up.

首先，我想告訴大家如何設置。

It is ridiculously easy and fast to run your own AI on your laptop, computer, or whatever it is.

在筆記本電腦、計算機或其他任何設備上運行自己的人工智能，簡單快捷得令人髮指。

This is free, it's amazing, it'll take you about five minutes.

這是免費的，非常神奇，只需五分鐘。

And if you stick around to the end, I want to show you something even crazier, a bit more advanced.

如果你堅持到最後，我想給你看一些更瘋狂、更先進的東西。

I'll show you how you can connect your knowledge base, your notes, your documents, your journal entries, to your own Private GPT, and then ask it questions about your stuff.

我將向你展示如何將你的知識庫、筆記、文檔、日誌條目與你的私人 GPT 相連接，然後向它提出有關你的東西的問題。

And then second, I want to talk about how Private AI is helping us in the area we need help most, our jobs.

其次，我想談談私人人工智能如何在我們最需要幫助的領域--我們的工作--幫助我們。

You may not know this, but not everyone can use ChatGPT or something like it at their job.

您可能不知道，並不是每個人都能在工作中使用 ChatGPT 或類似功能。

Their companies won't let them, mainly because of privacy and security reasons.

他們的公司不允許他們這樣做，主要是因為隱私和安全原因。

But if they could run their own Private AI, that's a different story.

但如果他們能運行自己的私人人工智能，那就另當別論了。

That's a whole different ballgame.

這完全是兩碼事。

And VMware is a big reason this is possible.

而 VMware 正是實現這一點的重要原因。

They are the sponsor of this video, and they're enabling some amazing things that companies can do on-prem in their own data center to run their own AI.

他們是本視頻的贊助商，公司可以在自己的數據中心運行自己的人工智能。

And it's not just the cloud, man, it's like in your data center.

不僅是雲計算，就連數據中心也是如此。

The stuff they're doing is crazy.

他們做的事情太瘋狂了。

We're gonna talk about it here in a bit.

我們稍後再談。

But tell you what, go ahead and do this.

這樣吧，你來做吧。

There's a link in the description.

說明中有一個鏈接。

Just go ahead and open it and take a little glimpse at what they're doing.

快打開看看他們在做什麼吧。

We're gonna dive deeper, so just go ahead and have it open right in your second monitor or something, or on the side, or minimize.

我們要深入研究，所以請在你的第二個顯示器或其他地方打開它，或者在側面，或者最小化。

I don't know what you're doing, I don't know how many monitors you have.

我不知道你在做什麼，也不知道你有多少臺顯示器。

You have three, actually, Bob.

其實你有三個，鮑勃

I can see you.

我看見你了

Oh, and before we get started, I have to show you this.

哦，在我們開始之前，我必須給你看這個。

You can run your own private AI that's kind of uncensored.

你可以運行自己的私人人工智能，這種人工智能是未經審查的。

Like, watch this.

比如，看這個。

I love you, dude, I love you.

我愛你 夥計 我愛你

So yeah, please don't do this to destroy me.

所以，是的，請不要這樣做毀了我。

Also, make sure you're paying attention.

此外，請務必集中注意力。

At the end of this video, I'm doing a quiz.

在視頻的最後，我要做一個小測驗。

And if you're one of the first five people to get 100% on this quiz, you're getting some free coffee.

如果您是前五名在測試中獲得 100% 高分的人之一，您將獲得免費咖啡。

Network Chuck coffee.

網絡查克咖啡

So take some notes, study up, let's do this.

所以，做一些筆記，好好學習，讓我們開始吧。

Now, real quick, before we install a private local AI model on your computer, what does it even mean?

現在，在我們在你的電腦上安裝私人在地人工智能模型之前，請快速瞭解一下，它到底意味著什麼？

What's an AI model?

什麼是人工智能模型？

At its core, an AI model is simply an artificial intelligence pre-trained on data we've provided.

人工智能模型的核心是根據我們提供的數據預先訓練的人工智能。

One you may have heard of is OpenAI's chat GPT, but it's not the only one out there.

您可能聽說過 OpenAI 的哈拉 GPT，但它並不是唯一的一種。

Let's take a field trip.

讓我們進行一次實地考察。

We're gonna go to a website called huggingface.co.

我們要去一個叫 huggingface.co 的網站。

Just an incredible brand name, I love it so much.

這真是一個不可思議的品牌名稱，我非常喜歡。

This is an entire community dedicated to providing and sharing AI models.

這是一個致力於提供和共享人工智能模型的完整社區。

And there are a ton.

而且有很多。

You're about to have your mind blown, ready?

你將大開眼界，準備好了嗎？

I'm gonna click on models up here.

我要點擊上面的模型。

Do you see that number?

你看到這個數字了嗎？

505,000 AI models.

50.5 萬個人工智能模型。

Many of these are open and free for you to use, and they're pre-trained, which is kind of a crazy thing.

其中很多都是開放的，你可以免費使用，而且它們都經過預先培訓，這有點瘋狂。

Let me show you this.

讓我給你看看這個。

We're gonna search for a model named Llama 2, one of the most popular models out there.

我們將搜索一款名為 "Llama 2 "的機型，它是最受歡迎的機型之一。

We'll do Llama 2 7B.

我們將進行喇嘛 2 7B。

I, again, I love the branding.

我還是喜歡這個品牌。

Llama 2 is an AI model known as an LLM or large language model.

Llama 2 是一種人工智能模型，被稱為 LLM 或大型語言模型。

OpenAI's chat GPT is also an LLM.

OpenAI 的哈拉 GPT 也是一名法學碩士。

Now this LLM, this pre-trained AI model was made by Meta, AKA Facebook.

現在這個 LLM，這個預先訓練好的人工智能模型是由 Meta（又名 Facebook）製作的。

And what they did to pre-train this model is kind of insane.

他們對這個模型的預先訓練簡直是瘋了。

And the fact that we're about to download this and use it, even crazier.

而事實上，我們正要下載並使用它，這就更瘋狂了。

Check this out.

看看這個

If you scroll down just a little bit, here we go, training data.

如果你向下滾動一點，就會看到訓練數據。

It was trained by over 2 trillion tokens of data from publicly available sources, instruction data sets, over a million human annotated examples.

它由超過 2 萬億個來自公開來源的數據、指令數據集和超過一百萬個人工註釋示例進行訓練。

Data freshness, we're talking July, 2023.

數據新鮮度，我們說的是 2023 年 7 月。

I love that term, data freshness.

我喜歡 "數據新鮮度 "這個詞。

And getting the data was just step one.

獲取數據只是第一步。

Step two is insane because this is where the training happens.

第二步是瘋狂的，因為培訓就在這裡進行。

Meta, to train this model, put together what's called a super cluster.

為了訓練這個模型，Meta 組建了一個超級集群。

It already sounds cool, right?

聽起來已經很酷了，對吧？

This sucker is over 6,000 GPUs.

這傢伙的 GPU 超過 6000 個。

It took 1.7 million GPU hours to train this model.

該模型的訓練花費了 170 萬 GPU 小時。

And it's estimated it costs around $20 million to train it.

據估計，它的訓練費用約為 2000 萬美元。

And now Meta's just like, here you go kid, download this incredibly powerful thing.

現在 Meta 就像在說，給你，孩子，下載這個無比強大的東西。

I don't want to call it a being yet.

我還不想稱它為存在。

I'm not ready for that.

我還沒準備好。

But this intelligent source of information that you can just download on your laptop and ask it questions.

但這個智能信息源，你只需下載到筆記本電腦上，就可以向它提問。

No internet required.

無需網絡。

And this is just one of the many models we could download.

這只是我們可以下載的眾多模型之一。

They have special models like text to speech, image to image.

它們有特殊的模式，如文本到語音、影像到影像。

They even have uncensored ones.

甚至還有未經審查的。

They have an uncensored version of Allama too.

他們還有未經審查的《阿拉瑪》版本。

This guy, George Sung, took this model and fine tuned it with a pretty hefty GPU, took him 19 hours and made it to where you could pretty much ask this thing anything you wanted.

這個叫喬治-宋的傢伙利用這個模型，並通過一個相當大的 GPU 對其進行了微調，他花了 19 個小時，終於做到了你可以問這個東西任何你想要的東西。

Whatever question comes to mind, it's not going to hold back.

無論想到什麼問題，它都不會退縮。

So how do we get this fine tuned model onto your computer?

那麼，我們如何把這個經過微調的模型放到你的電腦上呢？

Well, actually I should warn you, this involves quite a bit of Allamas, more than you would expect.

實際上，我得提醒你，這裡面涉及到不少阿拉瑪斯，比你想象的要多。

Our journey starts at a tool called Allama.

我們的旅程從一個名為 "阿拉瑪 "的工具開始。

Let's go ahead and take a field trip out there real quick.

我們先去實地考察一下吧。

We'll go to allama.ai.

我們去 allama.ai。

All we have to do is install this little guy, Mr. Allama.

我們要做的就是安裝這個小傢伙，阿拉瑪先生。

And then we can run a ton of different LLMs.

然後，我們就可以運行大量不同的 LLM。

Llama2, Code Llama, told you lots of llamas.

拉瑪 2，代號 "拉瑪"，告訴你有很多拉瑪。

And there's others that are pretty fun like Llama2 Uncensored, more llamas.

還有一些其他的遊戲也很有趣，比如《Llama2 Uncensored》，裡面有更多的駱駝。

Mistral, I'll show you in a second.

米斯特爾，我馬上就給你看。

But first, what do we install Allama on?

但首先，我們要在什麼上面安裝阿拉瑪？

We can see right down here that we have it available on Mac OS and Linux, but oh, bummer, Windows coming soon.

我們可以看到，我們已經在 Mac OS 和 Linux 上提供了該軟件，但無奈的是，Windows 即將推出。

It's okay, because we've got WSL, the Windows Subsystem for Linux, which is now really easy to set up.

沒關係，因為我們已經有了 WSL，Linux 的 Windows 子系統，現在設置起來非常簡單。

So we'll go ahead and click on download right here.

是以，我們將繼續點擊下載。

For Mac OS, you'll just simply download this and install it like one of your regular applications.

對於 Mac OS，只需下載並安裝即可，就像安裝普通應用程序一樣簡單。

For Linux, we'll click on this.

對於 Linux，我們將點擊這個。

We got a fun curl command that will copy and paste.

我們有一個有趣的 curl 命令，可以複製和粘貼。

Now, because we're going to install WSL on Windows, this will be the same step.

現在，因為我們要在 Windows 上安裝 WSL，所以這將是相同的步驟。

So, Mac OS folks, go ahead and just run that installer.

是以，Mac OS 的用戶請繼續運行安裝程序。

Linux and Windows folks, let's keep going.

Linux 和 Windows 的朋友們，讓我們繼續。

Now, if you're on Windows, all you have to do now to get WSL installed is launch your Windows terminal.

現在，如果你使用的是 Windows 系統，要安裝 WSL，只需啟動 Windows 終端即可。

Just go to your search bar and search for terminal.

只需在搜索欄中搜索終端。

And with one command, it'll just happen.

只要一聲令下，它就會發生。

It used to be so much harder, which is WSL dash dash install.

以前的安裝難度很大，這就是 WSL 儀表盤的安裝。

It'll go through a few steps.

它會經過幾個步驟。

It'll install Ubuntu as default.

它將默認安裝 Ubuntu。

I'll go ahead and let that do that.

我會繼續讓它這樣做。

And boom, just like that, I've got Ubuntu 22.04.3 LTS installed, and I'm actually inside of it right now.

就這樣，我安裝好了 Ubuntu 22.04.3 LTS，現在我就在裡面。

So now at this point, Linux and Windows folks, we've converged, we're on the same path.

是以，在這一點上，Linux 和 Windows 的朋友們，我們已經趨於一致，走在了同一條道路上。

Let's install Olama.

讓我們安裝 Olama。

I'm going to copy that curl command that Olama gave us, jump back into my terminal, paste that in there, and press enter.

我要複製 Olama 給我們的 curl 命令，跳回到我的終端，把它粘貼進去，然後按回車鍵。

Fingers crossed, everything should be going great, like the way it is right now.

希望一切順利，就像現在這樣。

It'll ask for my sudo password.

它會詢問我的 sudo 密碼。

And that was it.

就是這樣。

Olama is now installed.

Olama 已安裝完畢。

Now, this will directly apply to Linux people and Windows people.

現在，這將直接適用於 Linux 用戶和 Windows 用戶。

See right here where it says NVIDIA GPU installed?

看到這裡顯示已安裝 NVIDIA GPU 嗎？

If you have that, you're going to have a better time than other people who don't have that.

如果你有這種能力，你就會比其他沒有這種能力的人過得更好。

I'll show you here in a second.

我馬上就給你看。

If you don't have it, that's fine.

如果沒有，也沒關係。

We'll keep going.

我們將繼續前進。

Now let's run an LLM.

現在，讓我們運行一個 LLM。

We'll start with Llama 2.

我們從拉瑪 2 號開始。

So we'll simply type in Olama, run, and then we'll pick one, Llama 2.

是以，我們只需輸入 Olama，運行，然後選擇一個，Llama 2。

And that's it.

就是這樣。

Ready, set, go.

準備，準備，開始

It's going to pull the manifest.

它將拉動艙單。

It'll then start pulling down and downloading Llama 2, and I want you to just realize this, that powerful Llama 2 pre-training we talked about, all the money and hours spent, that's how big it is.

然後它就會開始下拉並下載 Llama 2，我希望你能意識到這一點，我們所說的強大的 Llama 2 前期培訓，所花費的所有金錢和時間，就是這麼大。

This is the 7 billion parameter model, or the 7B.

這就是 70 億參數模型，或稱 7B。

It's pretty powerful.

它非常強大。

And we're about to literally have this in the palm of our hands.

而這一切，就在我們的掌中。

In like three, two, one.

三 二 一

Oh, I thought I had it.

哦，我以為我拿到了。

Anyways, it's almost done.

總之，就快完工了。

And boom, it's done.

然後 "砰 "的一聲，就完成了。

We've got a nice success message right here, and it's ready for us.

我們這裡有一個很好的成功資訊，它已經為我們準備好了。

We can ask you anything.

我們可以問你任何問題。

Let's try, what is a pug?

讓我們試試，什麼是八哥犬？

Now, the reason this is going so fast, just like a side note, is that I'm running a GPU, and AI models love GPUs.

現在，這一切之所以進行得如此之快，就像一個附帶說明，是因為我正在運行 GPU，而人工智能模型喜歡 GPU。

So let me show you real quick.

所以，讓我給你演示一下。

I did install Llama on a Linux virtual machine.

我在 Linux 虛擬機上安裝了 Llama。

And I'll just demo the performance for you real quick.

我將為你快速演示一下性能。

By the way, if you're running like a Mac with an M1, M2, or M3 processor, it actually works great.

順便說一句，如果你使用的是裝有 M1、M2 或 M3 處理器的 Mac，它的運行效果會非常好。

I forgot to install it.

我忘記安裝了。

I gotta install it real quick.

我得趕緊安裝。

And it'll ask you that same question, what is a pug?

它會問你同樣的問題：什麼是八哥犬？

It's going to take a minute.

需要一分鐘

It'll still work, but it's going to be slower on CPUs.

它仍然可以工作，但 CPU 運行速度會變慢。

And there it goes.

就是這樣。

It didn't take too long, but notice it is a bit slower.

花的時間並不長，但注意到速度有點慢。

Now, if you're running WSL, and you know you have an Nvidia GPU and it didn't show up, I'll show you in a minute how you can get those drivers installed.

現在，如果你正在運行 WSL，而且你知道你有一個 Nvidia GPU，但它沒有顯示出來，我馬上就會告訴你如何安裝這些驅動程序。

But anyways, just sit back for a minute, sip your coffee, and think about how powerful this is.

但無論如何，請稍安勿躁，喝一口咖啡，想想這有多麼強大。

The tinfoil hat version of me, stinkin' loves this.

戴著錫箔帽的我非常喜歡這個。

Because let's say the zombie apocalypse happens, right?

因為假設殭屍啟示錄發生了，對嗎？

The grid goes down, things are crazy.

電網癱瘓，一切都瘋了。

But as long as I have my laptop and a solar panel, I still have AI, and it can help me survive the zombie apocalypse.

但只要我有筆記本電腦和太陽能電池板，我就仍然擁有人工智能，它可以幫助我在殭屍啟示錄中倖存下來。

Let's actually see how that would work.

讓我們來看看實際效果如何。

It gives me next steps.

它告訴我下一步該怎麼做。

I can have it help me with the water filtration system.

我可以讓它幫我安裝水過濾系統。

This is just cool, right?

這很酷吧？

It's amazing.

太神奇了

But can I show you something funny?

但我能給你看點有趣的東西嗎？

You may have caught this earlier.

你可能已經注意到了。

Who is Network Chuck?

誰是網絡查克？

What?

什麼？

Dude, I've always wanted to be Rick Grimes.

老兄，我一直想成為瑞克-格蘭姆斯。

That is so fun.

太有趣了

But seriously, it kind of like hallucinated there.

不過說真的，那裡有點像幻覺。

It didn't have the correct information.

它沒有正確的資訊。

It's so funny how it mixed the zombie apocalypse prompt with me.

它把殭屍啟示錄的提示和我混在一起，真是太有趣了。

I love that so much.

我非常喜歡。

Let's try a different model.

讓我們試試不同的模式。

I'll say bye.

我會說再見的。

I'll try a really fun one called Mistral.

我想試試一款非常有趣的產品，名叫 "米斯特拉爾"（Mistral）。

And by the way, if you want to know which ones you can run with OLAMA, which LLMs, they got a page for their models right here.

順便說一下，如果你想知道哪些可以用 OLAMA 運行，哪些是 LLM，他們在這裡有一個模型頁面。

All the ones you can run, including LLAMA2 Uncensored.

你能運行的所有程序，包括 LLAMA2 Uncensored。

Wizard Math.

魔法數學

I might give that to my kids, actually.

事實上，我可能會把它送給我的孩子們。

Let's see what it says now.

讓我們看看它現在怎麼說。

Who is Network Chuck?

誰是網絡查克？

Now my name is not Chuck Davis.

我現在不叫查克-戴維斯了。

And my YouTube channel is not called Network Chuck on Tech.

我的 YouTube 頻道也不叫 "科技網絡查克"。

So clearly the data this thing was trained on is either not up to date or just plain wrong.

是以，很明顯，這東西所依據的數據要麼不是最新的，要麼就是完全錯誤的。

So now the question is, cool, we've got this local private AI, this LLM.

所以現在的問題是，酷，我們已經有了這個在地的私人人工智能，這個法學碩士。

That's super powerful.

太強大了

But how do we teach it the correct information for us?

但是，我們該如何向它傳授正確的資訊呢？

How can I teach it to know that I'm Network Chuck, Chuck Keith, not Chuck Davis, and my channel's called Network Chuck?

我怎麼教它知道我是網絡查克、查克-基思，而不是查克-戴維斯，我的頻道叫網絡查克？

Or maybe I'm a business and I want it to know more than just what's publicly available.

或者，我是一家企業，我希望它知道的不僅僅是公開的資訊。

Because sure, right now, if you downloaded this LLM, you could probably use it in your job.

因為可以肯定的是，現在，如果你下載了這個法學碩士學位，你可能會在工作中用到它。

But you can only go so far without it knowing more about your job.

但是，如果它不對你的工作有更多瞭解，你也只能走這麼遠。

For example, maybe you're on a help desk.

例如，您可能在服務檯工作。

Imagine if you could take your help desk's knowledge base, your IT procedures, your documentation.

試想一下，如果你能將服務檯的知識庫、IT 程序和文檔都納入其中，你會怎樣做？

Not only that, but maybe you have a database of closed tickets, open tickets.

不僅如此，您可能還有一個已關閉票據和未關閉票據數據庫。

If you could take all that data and feed it to this LLM and then ask it questions about all of that, that would be crazy.

如果你能把所有這些數據輸入這個法學碩士，然後就所有這些向它提問，那就太瘋狂了。

Or maybe you want it to help troubleshoot code that your company's written.

或者，您可能希望它能幫助排除貴公司編寫的代碼的故障。

You can even make this LLM public facing for your customers.