Building GPT4ALL from source code

[t-dasun]

Can some one guide me to setup gpt4all from src code.

I have used this link
https://github.com/nomic-ai/gpt4all/blob/main/gpt4all-chat/build_and_run.md

I installed qt as in the description.
Build gpt4all-backend and the gpt4all-chat
cmake ../../gpt4all-backend/ -DLLMODEL_CUDA=OFF -DLLMODEL_KOMPUTE=OFF

So, if the executable is chat that generates in gpt4all-chat/build/ bin when it execute I'm getting following

constructGlobalLlama: could not find Llama implementation for backend: kompute
constructGlobalLlama: could not find Llama implementation for backend: cuda
[Warning] (Thu Jun 13 23:20:20 2024): QQmlApplicationEngine failed to load component
[Warning] (Thu Jun 13 23:20:20 2024): qrc:/gpt4all/main.qml:66:5: Type ChatView unavailable
[Warning] (Thu Jun 13 23:20:20 2024): qrc:/gpt4all/qml/ChatView.qml:483:5: Type SettingsDialog unavailable
[Warning] (Thu Jun 13 23:20:20 2024): qrc:/gpt4all/qml/SettingsDialog.qml:104:9: Type MySettingsStack unavailable
[Warning] (Thu Jun 13 23:20:20 2024): qrc:/gpt4all/qml/MySettingsStack.qml:67:30: IconLabel is not a type
[Warning] (Thu Jun 13 23:20:20 2024): QIODevice::read (QNetworkReplyHttpImpl): device not open
[Warning] (Thu Jun 13 23:20:20 2024): ERROR: Couldn't parse:  "" "illegal value"

Can someone direct me to proper document. That how can I use bindings and how to set the model that we need to run.

[cosmic-snow]

I'm not entirely sure what you're doing there.

At first glance, this looks like you're building in the backend folder separately, but what you should do is just build from gpt4all-chat/ if you want to have the GUI. Note that there is a CMakeLists.txt there. This one will automatically include a backend build, too.

If your goal is to script the whole compilation you could also have a look at how it's done in CI. Although I'd recommend to just use Qt Creator for starters, it's easier and it's documented.