Just let me know when you have concluded all your experiments doc and then I can pick one
Just let me know when you have concluded all your experiments doc and then I can pick one
ZHi! I ran into an issue with onetrainer can I send it here if anyone can help?
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=16)
Fyou can see testing completed ones here
Fsure what is it?
ZFollowing your stable diffusion guide, I had setup the vm and everything but just running into issues launching onetrainer (it is working on vm but not on my pc)
Z@Dr. Furkan Gözükara let me know
Finstall error
Fso i would suggest this
Fon your pc
Ffollow this tutorial
Fwithout skipping any part
Fafter that fresh install onetrainer
F99% will be fixed
FYouTubeSECourses
If you are interested in using AI, generative AI applications, open source applications in your computer, than this is the most fundamental and important tutorial that you need. In this tutorial I show and explain how to properly install appropriate Python versions accurately, how to switch between different Python versions, how to install diffe...
ZI did
Fin that case make a fresh install
Fand send cmd logs here
Zof everything?
Fyes
Fentire cmd
Fso we can see errors
Zas in python cuda and the rest of the tools?
Znot even sure how i can cleanly remove them
Fno install onetrainer
Fand installation cmd logs
Zsure
Fmake a fresh one delete old one
ZI did that once
Zoriginally i installed on e drive
Zthen didnt work so I installed on c
Zwill try again and send u the logs
Zthanks
Fyou are welcome
Zhow can i uninstall it? by just deleting the folder?
Dis there a mixmum grid size A1111/forge can do? ETA is more than one hour...
DI hope it doesn't fail at the end
Fyep
Fyes there is
Fbut it checks before you startr
Dthanks
Dthe kohya model splitting / offloading feature that you are using, is it with bf16? or does kohya even use a lower precision for full fine tune
Fnope it cant train fp8 yet
Ffine tuning
Fthat is how works
Falso on lora, it still train bf16 but model is converted into fp8 weights
Dfor lora it's easy to use a model in low precision, because the values you are changing are still higher precision. but for full finetune he uses less than 16bit?
Ffor fine tune only 16 bit
Fnot lesser
Dfor full finetune, those 8 bit are the weights that you are training