>>105776096
if you used the CPU version then it will ignore your GPU. I'm not sure what happens if you use cuda with an amd. afaik amd isn't supported but i don't actually know.
>you got it to run
did you deploy it on a system or an actual server container through the yaml script?
>default model doesn't work so well for my stuff
this is actually why I consider it a big deal and upgrade from femboy. Femboy seemed buttery smooth but it was only programmed to use the one model and is abandonware and i could never get anyone else to work on it, but this one allows you to use multiple models for tagging so if one is not working great you can add another.