Many readers mentioned that they would like toLocal deploymentofDeepSeek-The R1 has networking capabilities, so let's talk about implementing them today.
I. Deploying the model locally with Ollama (if already deployed, ignore this section)
Download Ollama via the URL: https://ollama.com.
Multiple versions for Windows, Linux, macOS are available:
Take Windows as an example, install it directly after downloading, the window will be closed automatically after completion, and then you can judge whether Ollama is started or not by the system tray icon:
Click on the "Models" button at the top left of the Ollama website and select DeepSeek-R1:
Go to this screen and click on the drop-down menu to select the appropriate version (the higher the number, the higher the video memory requirement):
The approximate correspondence between model version and video memory requirements:
We download the model for 14b, click here to copy the command: ollama run deepseek-r1:14b
Click on the window icon in the menu bar at the bottom of Windows, type "powershell" in "Search Applications, Settings, and Documents", and then click "Windows PowerShell":
In the pop-up window, paste the command you just copied: ollama run deepseek-r1:14b and enter to start the download.
When finished, the word "Send a message" appears, indicating success. You can close the window at this point.
Second, download page-assist plug-in
The plugin currently supports Chrome and Firefox better, and also Edge (manual installation required).
1. Chrome
Download at https://www.google.cn/intl/zh-CN/chrome
The official download requires magic, and while you can find downloads offered by third parties, the subsequent manipulation involves the Chrome Store, which still requires magic.
After installing Chrome, click on the three dots in the upper right corner and select "Extensions" and "Visit Chrome App Store":
Enter "page assist" in the search box of the app store to find the plugin we want to use and click on it:
Click "Add to Chrome":
A pop-up will appear, click "Add Extension":
Once added, page-assist can be found in the extensions at the top of your browser as a dialog bubble icon:
2、Firefox
Download at https://www.firefox.com.cn
No magic is required throughout, and Firefox is recommended for those who can't handle the internet.
After installing Firefox, click on the extension icon in the upper right corner and select "Manage Extensions":
Search for "page assist" at "Find more add-ons":
The first one that comes out is, click on it:
Select "Add to Firefox":
Once done, you can enable page-assist by clicking on the "Manage Extensions" tab:
3. Edge browser
You can't find page-assist in the Microsoft Extension Store, so you must install this plugin manually (magic required).
Open the command line window of "Windows PowerShell" and execute the following commands:
powershell -c "irm bun.sh/install.ps1 | iex"
git clone https://github.com/n4ze3m/page-assist.git
cd page-assist
bun install
bun run build
When it's done, you can see the directory build in the directory you cloned locally:
Click the three dots in the upper right corner of Edge and select "Extensions":
Select "Manage Extensions" in the pop-up window:
In this interface, click "Extensions" on the left, and click "Developer Mode" to turn it on, then an icon "Load Extension" appears on the right:
Click on it and find the chrome-mv3 subdirectory under the build directory you just created and click "Select Folder":
Then you have page-assist in your list of extensions, success.
III. Use of page-assist
No matter which browser you open page-assist through, the interface is the same and it will detect that you are running Ollama:
Click on the drop down menu above and select the model you have installed:
At the bottom of the lower dialog box is a globe symbol; clicking on it to open it opens the web search function:
Let's try to ask a question that requires up-to-date information, which is the result of not having networking turned on:
This is the result of turning on networking:
As you can see, the new information was indeed added inside the answer, with the source of the information at the bottom.
One more weather report:
Let's go ahead and make some settings on page-assist to get it in tip-top shape.
Click on this gear in the upper right corner:
Under this menu, change the "Language" column to "Simplified Chinese":
Go back to the main interface, you will find that the interface is already in Chinese:
Again, click on the gear in the upper right corner to enter the menu, scroll down to find "Manage Web Search", where you can change the search engine used. The default search engine DuckDuckGo doesn't work well in Chinese.Suggest to change to Google, Baidu, Sogou.
At this point, a powerful networkable DeepSeek is ready to serve you!
The article covers the URL:
page-assist code page:
https://github.com/n4ze3m/page-assist
Ollama's official website:
https://ollama.com