aa956

Same behavior. May this be related? https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/12822

It's hard to port Windows app to web, you'll have to basically scrap and re-implement most of the UI part. Even changing the GUI framework is hard in .NET world...

Actually it looked interesting and I've got breadboard running on debian. Don't know how to get rid of electron (so you need a desktop environment unless you know some nodejs...

Theoretically? Absolutely, app is just javascript web server running on node.js Problem is, it includes the [electron](https://www.electronjs.org/) (basically a full blown web browser) as a dependency to present itself as...

I've just installed it today taking notes: # Installation for AMD GPU-s on linux Tested on Debian 12 (bookworm) with RX 6700 XT. Used as desktop for some time so...

Same problem with locale having ',' as a decimal separator:  One more note: CSV is exported broken as both decimal and list separators used are `,` (comma), although as...

One more problem probably with the same cause: broken localization. Export of negative numbers to TSV results in "?" instead of "-" prefix for a number (sample data with x...

If I understand correctly [their training code is not available and is not planned to be released](https://huggingface.co/ByteDance/SDXL-Lightning/discussions/1) There's nothing to implement in OneTrainer then? Regarding using it, see this, maybe...

Hardlinks or symlinks will probably work? Don't forget to separately commit: * [Optimizer settings](https://github.com/Nerogar/OneTrainer/issues/116#issuecomment-1880036256) `training_user_settings/optimizer_prefs.json` * Concepts settings `training_concepts/concepts.json` * Sample settings too? Unfortunately I see no simple way to...

It may be a good idea to remove the check for system python version if conda is detected. Or at least first do `conda activate`? Maybe add a check for...