Panda has posted a lot of projects on how to build GPT on NAS before. And what I'm going to write today is the best private deployment robot framework I've ever used.,The project is provided by fans.,So in the future, if you have any projects that you are interested in but won't deploy.,You can send it to Panda's side.,Panda can go to study and toss.,Looking forward to more interaction with everyone.。
experience
The project integrates a lot of functions, and the many functions are also lightweight, and the project size is more than 300MB. At the same time, the deployment is also very simple,The overall UI of the interface may be a little messy,A single person seems to be at least he is very rich,The functions of each section are also very clear,The project address is not convenient for me to put here,Put a picture,You can go to github to search for a little star。
Project page
Log in and you can see a small assistant that comes with it,Same as the use of GPT,You can see the principle based on the model,Here is the 3.5 model,If you need 4.0,You can also find a way to get the key and fill it in the settings。
Main interface
Model adjustments
In addition to the regular GPT services, the most important thing is that he actually supports plug-ins!! which is amazing, through plug-ins we can achieve more functions, such as drawing, weather, intervention in search engines, etc., on the one hand, he makes up for the shortcomings of the 3.5 model, on the other hand, it also provides richer gameplay.
Plug-in marketplace
At the same time, the plugin also supports third-party customization, if you have the ability, you can edit the plugin yourself and access it to the project. Of course, there is one of the most interesting things about this project, that is, the character setting, you can determine your model attributes through the character setting, for example, you can let him play the little Lori, or the strange Shu Xiao, or even let him play the panda directly.
Character setting
Of course, if you're too lazy to set up a character edit character card, the project also provides some presets for you to use. You can choose from the second option in the left ribbon.
Preset models
Finally, it's the setting item, where you can set OpenAI's proxy and key, and of course, you can also set the theme of the project and so on. You can try it yourself.
Settings
deploy
The next step is our deployment process, here we need to prepare the API_KEY of OpenAI and the proxy address of OpenAI, which I mentioned in the previous article on deploying GPT how to obtain, you need to go to GitHub's project to get it, and here is a screenshot of the project.
Screenshot of the project
The permission requirements for the project are not high, so you can use any NAS that supports Docker to deploy it, and download the image by searching for the project: lobehub/lobe-chat. At the same time, we need to set some environmental parameters, OPENAI_API_KEY, OPENAI_PROXY_URL and ACCESS_CODE, of which the first two are needless to say, and the last one is the project access password. The docker run command is as follows:
docker run -d -p 3210:3210 \
-e OPENAI_API_KEY=sk-填你自己的 \
-e OPENAI_PROXY_URL=填你自己的 \
-e ACCESS_CODE=访问密码 \
--name lobe-chat \
lobehub/lobe-chat
If there is a conflict between the ports, remember to change them. After opening the SSH port, Synology can enter this command. Then we can go directly to the browserhttp://nasip+3210.
order
summary
Very nice project with a lot of features. The role setting function of the project is still quite interesting,You can try to tune out your own exclusive GPT assistant,Another is the plug-in library,This undoubtedly gives more possibilities to the 3.5 model,If there are usually more users who use GPT projects,I think this is the most perfect private deployment service at present,Wall crack recommended。
The above is all the content of this issue, if you think it's interesting or helpful to you, you might as well like the collection, and finally hope to get your attention, let's see you next time!