VM "AI" and internal server error .

[frankz]

Hello everyone, surely some of you use an AI-type proxmox VM.

I have a chatgpt compatible server and it works properly thanks to the variation or rather the obligatory use of the host CPU .

Unfortunately, I can't get some AI modules to work, which are downloaded from the VM but not loaded into memory.

The error is strict-origin-when-cross-origin .

I don't know if I should actually research the lawsuit on the node, so I try to ask the same .....

 

[Nollimox]

I don't know if I should actually research the lawsuit on the node, so I try to ask the same .....

You lost me here with lawsuit...
I am actually working on this currently and had hope to use RancherOS. But, support/development ended; so, I'll use Debian minimal with AutoGPT or create my own. What OS are you using on your VM? Do you have an API key for all modules?

 

[t.lamprecht]

I'm not really understanding your post here completely, but can give you some hint's w.r.t. quoted error message:

The error is strict-origin-when-cross-origin .

See https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS (it is a bit complicated in the details though)

Basically this is a security protection of modern browsers to avoid that some scripts or attacks can easily leak your browser session state (login ticket-cookies and other things) to some other host (origin) that the current web-site/web-app is not related too.

Either ensure your backend tells the browser to allow request to your other origins that are used here (e.g., some API of some AI thingy), or route them all to the same backend, e.g., use nginx/caddy/... reverse proxying.

 

[frankz]

Sorry, the post is a little ambiguous or poorly explained. I have an ubuntu VM with a server downloaded from Github ( serge ) , with which I run a chatgpt server via docker . It always works (with the obligation to use the host cpu on proxmox), but I encountered some problems in loading some models into memory. I thought I would find some suggestions on the proxmox forum ( if it is necessary to modify some parameters beyond the host cpu). I think the problem is different, as being not very proficient on docker, I probably misconfigured the updates.

 

[frankz]

You lost me here with lawsuit...
I am actually working on this currently and had hope to use RancherOS. But, support/development ended; so, I'll use Debian minimal with AutoGPT or create my own. What OS are you using on your VM? Do you have an API key for all modules?

Ubuntu 22.04 lts (VM no LXC ) .

 

[Nollimox]

Ubuntu 22.04 lts (VM no LXC ) .

Okay, do you have API keys?

 

[frankz]

Okay, do you have API keys?

no

 

[Nollimox]

no

Well, I am not sure how you're doing yours; however, I know that each search modules requires an API key...unless you're just adding a desktop environment to access ChatGPT by a web browser.

 

[frankz]

Well, I am not sure how you're doing yours; however, I know that each search modules requires an API key...unless you're just adding a desktop environment to access ChatGPT by a web browser.

https://github.com/serge-chat/serge

 

[Nollimox]

https://github.com/serge-chat/serge

Interesting...thank you for sharing!

 

[frankz]

Interesting...thank you for sharing!

Very interesting, I recommend you download nouse-hermes and stable-vicuna . Also, if you install the server, let me know if when you refresh the available models, a new list will appear with a few but very recent modules. It's these modules that I can't load into memory but no problem for downloading. Also just today I downloaded the lxc ubuntu 22.04 template on proxmox and installed serge . It works prefettamete apart from what I wrote to you. If you can solve it, let me know, I want to solve this problem!

 

[gptnederlands_nl]

Hello everyone, surely some of you use an AI-type proxmox VM.

I have a chatgpt compatible server and it works properly thanks to the variation or rather the obligatory use of the host CPU .

Unfortunately, I can't get some AI modules to work, which are downloaded from the VM but not loaded into memory.

The error is strict-origin-when-cross-origin .

I don't know if I should actually research the lawsuit on the node, so I try to ask the same .....

I am similar to you, there is a compatibility with ChatGPT but still have some problems that need to be resolved.

 

[frankz]

I am similar to you, there is a compatibility with ChatGPT but still have some problems that need to be resolved.

What kind of problems do you have with chatgpt? What do you use for chatgpt ?