What is the main goal of the GPT4All project ?

[SuperUserNameMan]

Hello,

When I discovered GPT4All, I thought the main goal of the project was to create a user-friendly frontend UI to talk to local LLM. I thought the main project was the "Desktop Chat Client" displayed on the homepage.

But when I look at the project's github, I often see mentions of bindings and API and server mode.

Also, compared to other similar projects, the "Desktop Chat client" gives me the feeling that it is evolving relatively slowly, like if it was a secondary or ternary side-project.

Indeed, some of you might consider that if i'm not happy with GPT4ALL, i could use another project or contribute to the code. And I agree. However, the GPT4ALL desktop chat client has something that the other projects I tried don't have : simplicity, stability (i'm on linux) and a certain je ne sais quoi that make it difficult to replace for me. And if i had more free time, of course and indeed i'd certainly dare diving into the code to submit some pull requests. But I don't have that time and energy currently, so i'm just a user.

So, please take no offense in my questions, but :

• What evolution should I expect from the GPT4ALL "Desktop Chat Client" ?
• Will it become a real desktop tool or will it remain a simple basic frontend for I duno what's behind the scene ? (API, bindings, server mode, ...)
• Will it allow to manage chat sessions like we can manages files in other desktop applications ? (open/save/save as...)
• Will it become multi-modal like the basic llama.cpp server which is already compatible with llava ?

Or in one question : is it really useful that I keep posting "QoL feature requests" ?

[manyoso]

• What evolution should I expect from the GPT4ALL "Desktop Chat Client" ?

We are planning a major new release with a completely changed UI theme at the end of this month. We will be releasing today offline installers (and nightly versions) that track the feature branch where this work is being done.

• Will it become a real desktop tool or will it remain a simple basic frontend for I duno what's behind the scene ? (API, bindings, server mode, ...)

Not sure what you have in mind with 'a real desktop tool', but yes, we plan on major new features and polish of existing features that I hope will fit your definition :)

• Will it allow to manage chat sessions like we can manages files in other desktop applications ? (open/save/save as...)

This is already there, right? I'm not sure what you are asking because to my mind this already exists :)

• Will it become multi-modal like the basic llama.cpp server which is already compatible with llava ?

That's the plan! Notice today we just put out a new state of the art multimodal embedding model: https://blog.nomic.ai/posts/nomic-embed-vision

[SuperUserNameMan]

Wow, i'm all excited already :-D

Regarding the "open / save / save as..." chats session feature i was mentioning, I'm not sure if you're referring to the current 2.8.0 version when you say it is already there, but just in case : I was thinking more about a "File" menu like most desktop application have, and less of the current ChatGPT way of doing (side bar with list of all previous chats all stored into the same directory), so it would be possible to actually save and open filename.chat files in any directory of our choice.

IMHO, this kind of open / save / save as ... feature would make GPT4ALL "a realer desktop tool" because it could smoothly integrate into a multi-application desktop workflow.
For instance, if I use GPT4ALL to work on a project which involves multiple documents from different applications, I could store the related filename.chat sessions in the same directory as all the other files of the project.
Double clicking on the filename.chat could open it.
I could also move the filename.chat from a computer to an other, or backup it, etc, like any other file.

I hope I was clearer in my description, and that I'm not putting too much pressure on your shoulders.

Anyway, I can't wait to test the next version.

And thank you for your answer and for GPT4ALL ! :-)

[SuperUserNameMan]

Not sure what you have in mind with 'a real desktop tool', but yes, we plan on major new features and polish of existing features that I hope will fit your definition :)

@manyoso

First, 3.0.0 is a great progress. 👍

However, here is what prevents it from being a "real desktop tool" in my pragmatic user opinion : #1999

When you use a "real desktop tool", you can't wait for 5 or more minutes after you click quit before the application actually close itself.

We have the ability to create and save multiple chat sessions (edit: with the "save chat context" option enabled), but as soon as we have more than a few, it takes years to load and re-save ALL contextualized chats session, because they are ALL deserialized and reserialized one by one even if they are not going to be read or modified.

Instead of deserializing them all on startup : why not deserialize them on demand - ie only if the chat session is selected ?

Instead of re-serializing them all on quit : why not re-serialize and re-save only if modified ?