Name: ComfyUI 教學系列：EP13 - 探索 Ollama、LLaVA、Gemma 模型！ (ComfyUI Tutorial Series: Ep13 - Exploring Ollama, LLaVA, Gemma Models)
Uploaded: 2025-11-21T18:53:35.000Z
Duration: 19 min 54 s

Welcome to episode 13 of our ComfyUI tutorial series.

歡迎來到我們 ComfyUI 教學系列的第 13 集。

Today I want to show you how to install and run the Ollama tool and how you can test different large language models inside Comfyui with its help.

今天我要教大家怎麼安裝和執行 Ollama 這個工具，以及如何透過它在 Comfyui 裡測試不同的大型語言模型。

程度副詞

This way you can generate prompts from your instructions or even from images.

這樣你就可以從你的指令，甚至從圖片來生成提示。

First, we need a tool to manage all those large language models, so go to the ollama.com website.

首先，我們需要一個工具來管理所有這些大型語言模型，所以請到 ollama.com 這個網站。

As you can see, it can manage all kinds of models like Llama, Phi, Mistral and Lava.

你可以看到，它可以管理 Llama、Phi、Mistral 和 Lava 等各種模型。

If you want to learn more about it, there's a Docs section that links to the GitHub page.

如果你想了解更多，這裡有 Docs 區塊，裡面連結到 GitHub 頁面。

You'll also find more links here for a quick start and additional resources.

你也會在這裡找到更多 quick start 和額外資源的連結。

Back on the Ollama site you have the Download button where you can choose your operating system.

回到 Ollama 網站，你會看到 Download 按鈕，你可以選擇你的作業系統。

Mine is Windows, so I'll click Download for Windows.

我的作業系統是 Windows，所以我會點擊 Download for Windows。

I'll save it on my desktop for now and let's see how we install it.

我會先把它存在我的桌面，然後我們來看看怎麼安裝。

Just double click on it, then click Install and wait for it to finish.

直接雙擊它，然後點擊 Install 並等待它完成。

Once it's installed, you can see the Llama icon in the notification area that indicates it is running.

安裝完成後，你可以在通知區域看到 Llama 圖示，表示它正在運行。

If you want to close Alma, right, click there and choose Quit.

如果你想關閉 Alma，右鍵點擊那裡並選擇 Quit。

It's important to know this as it can use up some VRAM so you can close it when you're not using it.

知道這一點很重要，因為它會佔用一些 VRAM，所以不用時可以關閉它。

To start Ollama again, find it in your apps by clicking on Start and searching for Ollama.

要重新啟動 Ollama，請點擊開始並搜尋 Ollama 來在你的應用程式中找到它。

Now if I click on it, it might seem like nothing happened, but

現在如果我點擊它，可能會感覺什麼都沒發生，但

Ollama has started as we can see the icon with the Ollama in the notification area.

Ollama 已經啟動了，因為我們可以在通知區域看到帶有 Ollama 的圖示。

Back on the Ollama website, we have a section for models.

回到 Ollama 網站，我們有一個模型區塊。

If I click on it, we can find different models here such as Llama, Gemma, Mistral and so on.

如果我點擊它，我們可以找到不同的模型，例如 Llama、Gemma、Mistral 等等。

The top search is a little better than the filter below.

上方的搜尋比下方的篩選器好一點。

So if I search for Vision, for example, I can find models that can interpret an image and provide a description or caption for it.

所以如果我搜尋 Vision，例如，我可以找到能解讀圖片並提供描述或說明文字的模型。

If I search for Gemma, I can see different variations of that model.

如果我搜尋 Gemma，我可以看到這個模型的不同變體。

If I go back to the models, we can also filter them by newest or popular.

如果我回到模型頁面，我們也可以依據最新或熱門來篩選。

Let's say I want to install and test Gemma.

If I click on it and scroll down, I can read more information about it.

如果我點擊它並向下捲動，我可以閱讀更多關於它的資訊。

Here you can see different versions, but to see all, you need to click on view more.

這裡可以看到不同的版本，但要看全部，你需要點擊查看更多。

As you can see, it has all kinds of sizes.

你可以看到，它有各種不同的尺寸。

Usually the larger models are smarter, but it also depends on how many parameters they are trained on.

通常模型越大越聰明，但這也取決於它們訓練了多少參數。

For example 2B means 2 billion parameters.

So 7B has 7 billion parameters, which allows it to handle more complex tasks, but

所以 7B 有 70 億個參數，這讓它能處理更複雜的任務，但

If I select the 7B version, I see a command on the right to install that model.

如果我選擇 7B 版本，右邊會出現一個安裝該模型的指令。

If I choose a different model like to be instruct Q4, the command changes.

如果我選擇另一個模型，像是 instruct Q4，指令就會改變。

I can click on this icon to copy that command.

我可以點擊這個圖示來複製該指令。

Then I click Start, type cmd and select Command Prompt.

然後我點擊開始，輸入 cmd 並選擇「命令提示字元」。

Here I can paste the command I just copied and press Enter.

在這裡我可以貼上剛才複製的指令並按下 Enter。

This will pull all the information and models it needs to run.

這會下載所有需要的資訊和模型來運行。

This model has less than 2 gigabytes, so it should run quickly, but

這個模型不到 2GB，所以應該會跑得很快，但是

However, there is also a Gemma version 2 that seems to be better.

不過，也有一個 Gemma 2 版本，看起來更好。

Once the model finishes installing, we get a prompt asking us to send a message as suggested.

模型安裝完畢後，會出現一個提示，要我們照著建議傳送訊息。

You can type a message to get help and it gives some command suggestions like buy for exit.

你可以輸入訊息來尋求協助，它會提供一些指令建議，像是輸入 buy 來離開。

The help command also works to show all those commands we can communicate with the model right here.

help 指令也能顯示所有我們可以跟模型溝通的指令。

So if I type hi and hit enter, it will respond.

所以如果我打 hi 並按下 enter，它就會回應。

It's nice to interact with the model, but the interface is kind of basic.

跟模型互動感覺不錯，但介面有點陽春。

There are graphical interfaces like Open Web UI, but

有像 Open Web UI 這樣的圖形介面，但是

today I'll focus on using it inside ComfyUI I'll use by to exit.

今天我會著重在 ComfyUI 裡面使用它，我會用 by 來離開。

If you use the command alma list, you can see all the models you have installed.

如果你用 ollama list 這個指令，可以看到你安裝的所有模型。

As you can see, I have a few gigabytes here which can take up a lot of space on your hard drive.

你可以看到，我這裡有幾個 GB，這可能會佔用你硬碟很多空間。

So if you want to remove models you don't use anymore, select the name of the model you want to remove, like Stable Prompt, for example.

所以如果你想移除你不再使用的模型，就選取你想移除的模型名稱，例如 Stable Prompt。

Then use Control C to copy that name, type alma followed by rm, which stands for remove.

然後用 Control C 複製那個名稱，輸入 ollama 後面加上 rm，這是 remove 的縮寫。

If I run the list again, you can see it's not in the list anymore.

如果我再執行一次 list，你就可以看到它已經不在清單裡了。

I'll close this window and now let's go to comfyui.

我會關閉這個視窗，然後我們來去 ComfyUI。

Let's go to the manager and then to Custom nodes.

我們去 manager，然後到 Custom nodes。

Search for Ollama and install the comfyui ollama node by this user.

搜尋 Ollama，然後安裝這個使用者提供的 comfyui ollama node。

然後你就可以點擊 Restart 然後確定。

We can close this tab now since a new window will open.

我們可以關閉這個分頁了，因為會有一個新視窗打開。

Then double click on the canvas and search for alima.

然後在畫布上雙擊並搜尋 alima。

This one has a text output called response and we can add a Show any node from Comfy Easy.

這個節點有一個叫做 response 的文字輸出，我們可以從 Comfy Easy 加入一個 Show any node。

Here you can add your instructions or text.

在這裡你可以加入你的訓示或文字。

For example, if I say hi and run it, I get this response.

舉例來說，如果我打個招呼然後執行，我就會得到這個回應。

This is using the URL that points to Ollama running in the background.

這是使用指向背景執行的 Ollama 的 URL。

If I copy this link and paste it in the address bar, you can see it says ALMA is running.

如果我複製這個連結並貼到網址列，你會看到它顯示 ALMA 正在執行。

So we need ALMA to be running to use the models.

所以我們需要 ALMA 正在執行才能使用模型。

You can give all kinds of instructions like in ChatGPT.

你可以給予各種訓示，就像在 ChatGPT 一樣。

It depends on how smart the model is and how many billion parameters it has.

這取決於模型的聰明程度以及它有多少十億個參數。

But it's free and you can choose a model that fits your computer's power, allowing us to generate prompts with it.

但是它是免費的，而且你可以選擇一個適合你電腦效能的模型，這樣我們就可以用它來產生提示。

Let's delete the ALMA generate node and add another one called ALMA Generate Advanced.

讓我們刪除 ALMA Generate 節點，然後加入另一個叫做 ALMA Generate Advanced 的節點。

Connect the response output to the Show Any node.

連接 response output 到 Show Any 節點。

Don't use the context output as that will give a bunch of numbers.

不要使用 context output，因為那會得到一堆數字。

Now here we have the instructions for what we want to do, and

在頂部我們可以放我們的提示。

When we run this, it will interpret that prompt using the instructions.

當我們執行這個時，它會利用說明來解讀那個提示。

You can try all kinds of instructions to see what works best for you.

你可以試試看各種說明，看看哪個最適合你。

I also tried generating some instructions with Chat GPT.

我也試著用 Chat GPT 產生了一些說明。

I will post these long instructions for generating a prompt that works well most of the time, though it depends on the model.

我會貼上這些長說明，用來產生一個大部分時間都有效的提示，不過這還是取決於模型。

When I was trying to get the prompt to start with the type of image, I mentioned watercolor to see if it would start with that.

當我試著讓提示從圖片類型開始時，我提到了水彩，看看它會不會從那裡開始。

However, I need to try it with some larger models for better results or to play with settings.

不過，我需要用一些較大的模型來試試，以獲得更好的結果或調整設定。

Let's try this version to see if it works better.

我們來試試這個版本，看看會不會比較好。

From here you can take for example the 7B version, which is better, or look at Gemma version 2.0 with 7 billion parameters.

從這裡你可以選像是 7B 版本，它比較好，或是看看 Gemma 2.0 版本，它有 70 億個參數。

But just to show you how to add another model, I'll demonstrate with this, which I think is similar to the Q4 version, though I'm not sure.

不過，為了示範如何新增另一個模型，我會用這個來示範，我覺得它跟 Q4 版本很像，雖然我不確定。

Anyway, copy the command, go to start, search for cmd and open the command window.

反正，複製指令，然後到開始功能表，搜尋 cmd 並開啟命令提示字元。

Paste the command and wait for it to install.

Then close it with the command by or just close the window.

然後用指令關閉，或是直接關閉視窗。

Now back in ComfyUI it doesn't appear here.

現在回到 ComfyUI，它還沒出現在這裡。

If I try to refresh I get undefined model.

如果我試著重新整理，會出現 undefined model。

So you can either restart comfyUI, add the node again, or right click and choose Fix Node, which will recreate that node with default values.

所以你可以重啟 ComfyUI，重新新增節點，或是按右鍵選擇 Fix Node，這會用預設值重新建立那個節點。

There are ways to make it more accurate if you want to experiment.

如果你想實驗，有辦法讓它更精確。

字幕列表影片播放

ComfyUI 教學系列：EP13 - 探索 Ollama、LLaVA、Gemma 模型！ (ComfyUI Tutorial Series: Ep13 - Exploring Ollama, LLaVA, Gemma Models)

description

positive

accurate

interpret