Use a model remotely on a local network - use the client via the API or even spin up a webpage?

[accessiblepixel]

HI,

First off, thank you for the work on the project so far I like having the ability to play with some of the LLMs offline and locally in a very accessible manor... However I have few systems that are best suited to the task of running gpt4all - I've seen that there's a http API available, is it possible to have a centralised system that has the models and does the processing, and gives the results back to a remote client/web page?

Then I can host the 'processor' on a central machine and then access it from more underpowered hardware.

Thanks!

[cosmic-snow]

There are some ways, but only indirectly.

• You can activate the chat client's web API server, but it's localhost. To make it available to other machines you'd have to either edit the source code or set up some kind of reverse proxy (i.e. a different server which forwards the requests)
• There are language bindings through which you could expose it over the network, but you'd need to know how to program with these.
• There is a simple CLI based on the Python bindings, so if you have e.g. an SSH connection, you could access that through a terminal.
• Some kind of remote desktop / screen sharing software would be another option.
• I guess there are some projects independent from GPT4All itself which are building on top of the bindings. E.g. I know that one of the authors of the Typescript bindings is working on something, although that's still in an early stage.

[mastercracker3d]

Hi, i am working in a fork for using network for processing, please see meshgrid_gpt www.mc3d.cl