How do I start the gui with commandlinearguments in a notebook?

[ghost]

Im using a notebook to launch this repo on rented GPUs. So far this is working. I can do batches bigger than 1 but after generation the GUI freezes and I have to rerun the notebook cell. I want to generate images in very large resolutions, and this only works above 1024x1024 locally with the commandline argument --medvram

When I want to render images above 1024x1024 (sometimes it works up to 1280x1280) on a rented GPU they will be completely black for some reason.

I want to enable the commandline argument --medvram in my notebook but I have no idea where to put it? This command launches the UI via gradio:

%cd "{repo_storage_dir}/stable-diffusion-webui" !python webui.py --gradio-debug --share

Putting --medvram beside the gradio args does nothing. Where can I put the commandline argument so I can get it to work?

[ClashSAN]

Link to notebook you're using? Can you modify a "webui-user" .sh or .bat file?

[ghost]

I dont use Colab, I use this: https://proximacentaurib.notion.site/A-guide-to-getting-started-with-the-paperspace-port-of-AUTOMATIC1111-s-web-UI-for-ppl-who-get-nervou-b83c2213f17e452e8b0e37ba64fe9758

https://github.com/Engineer-of-Stuff/stable-diffusion-paperspace

I can modify the files, but I dont know which one is the right one. I tried putting --medvram into launch.py at the top but that doesnt work. The .bat file isnt used by the server, I think

[ClashSAN]

ok, so its adapted from some other notebook shown in pic
https://raw.githubusercontent.com/Engineer-of-Stuff/stable-diffusion-paperspace/main/docs/assets/final.png
you can ask the paperspace notebook maintainer where to set parameters too

[ghost]

Thanks I asked there but I solved the problem. For whatever reason, putting it beside the gradio args does seem to work. I dont know why it generated a black image when I tried it. Thanks for your help and looking into it : - )