"Kernel [...] is for sm80-sm100, but was built for sm37" Error

[pastanagafregida]

Hi! I've installed Stable Diffusion XL with AUTOMATIC1111 by following a tutorial, but every image I've created looks the same - a red-purple-blue-ish abstract pattern. And there's an error in the CMD window that says: FATAL: kernel 'fmha_cutlassF_f16_aligned_64x64_rf_sm80' is for sm80-sm100, but was built for sm37.

Below it there's this message, in case it's relevant to the issue (I don't understand what it means): A tensor with all NaNs was produced in VAE. Web UI will now convert VAE into 32-bit gloat and retry. To disable this behavior, disable the 'Automatically revert VAE to 32-bit floats' setting. To always start with 32-bit VAE, use --no-half-vae commandline flag.

I've searched through the Internet and it's seemingly related to a... plugin? called Xformers. But the values are different in each message error. Do you know what may the issue be and how can it be solved? I can't find anything named like that in the Stable Diffusion folder. Should I download Xformers? Which version? Is there anything else I should take into account to avoid messing things further? I'd really appreciate any help, I have no idea of computing and I don't know what I'm doing...

[missionfloyd]

I don't know what tutorial you followed, but I'd start over with the relevant setup guide for your hardware here:
https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki