Edna
Edna
Latest release doesn't have a fix for this, it's still an issue
Please note this will error out if the eos_token_id is a list (like in some llama3 configurations) 
In some configurations running llama, the eos_token_id is set to a list.  this works fine for sampling until prompt lookup decoding is added. It's a simple fix, though, adding...
Chroma PR is pretty close to done in https://github.com/huggingface/diffusers/pull/11698 I believe Nunchaku support should be easy from there? @Aprilhuu Take note of https://github.com/huggingface/diffusers/pull/11698#issuecomment-2969717180: distilled_guidance_layer may or may not play well...
Quantization shouldn't rely on the attention mask being correct (unless training, which I don't believe this quantization method does)
Ah good to know. I'll work on getting the attention mask working as soon as possible
This sounds unrelated since it is after the launch process has occurred (the zen browser opens), unless the browser is launching in the launch process and also hangs for ~2...
~~Switching from X11 -> Wayland fixes this (no idea why)~~ maybe not?
Issue is still present on the latest version
unsure if this still happens. I no longer can reproduce the issue but my system has changed a lot since then (including that im now on the flatpak vesktop instead...