-
-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Can't train a lora for SDXL from a fresh install. #2396
Comments
looks like something wrong with your CUDA installation. installed requirements.txt ? |
Did you install CUDA 11.8 as per the README instructions? |
I installed not only Kohya from the setup.bat, but also CuDDN, bitsandbytes, CUDA 11.8 (cuda_11.8.0_522.06_windows to be precise), the files required in the sd-script folder, and I updated PIP from venv and redid all of that...I'm not sure what else I need to do. |
I'm having a similar issue. I've been trying to get kohya to work for a few days, and I see a tangentially related error:
Same as @Deejay85 - CUDA 11.8, CuCDDN... ran everything from setup.bat. |
I was hoping that, since someone else is having the same problem I am, that an answer would be forthcoming, but apparently not. I did try to install the new version that was released today, and even tried to reinstall all the packages for Kohya just in case, but that didn't fix it either...I seem to be getting the same messages as I did before. I really don't know if it is something on my end, or if it's something on Kohya's end, but I really would wish the dev would take a look at it, so that I at least know where to start in fixing the problem. |
Didn't want to make a new thread, so I decided to bump the old one I made two weeks ago. I tried copying the newest release into Kohya, but that didn't make any difference, so even after two releases, I'm still having the same problems I did before. |
I ended up uninstalling anything related to python, cuda, nvidia, and microsoft development (cpp redistributables), then reinstalled and it fixed all of my issues. Before I also had Cuda 11.8 and 12.x installed and I'm guessing something went stupid there. So I stuck with cuda 11.8 this time. But not really sure - basically uninstalling and reinstalling fixed everything. |
Yeah, so many thing. A break down within the software stack… This was the best thing to do. Glad it fixed thing for you. |
I am having the exact same issue. Did anyone find a solution that does not involve reinstalling everything? |
I uninstalled everything as listed by Machineminded, and mine is still producing the same exact problems as before. Should I paste the entire log just to verify? |
Downloaded the newest version of Kohya, did a fresh install to a new directory, installed everything, and here are the results I got when I tried to train something:
In short the same old song and dance. 😩 Any advice? |
Not sure what happened but you should be able to install torch+cu118 and resolve this. Check this link: |
@Deejay85 The GPU IDs option on your config seems to be junk text. (It's an option located under the Accelerate launch category) |
I might add an input validator and log a message if it does not match the expected pattern |
I tried leaving it blank, with spaces, dashes, and as two blocks of text separated only by a hyphen...none of that worked. I am using only one graphics card BTW, because 4090s don't grow on trees you know? 😜 |
Supposing you kept it blank for GPU ID, does the log still show this error you had before like |
Same error. If you want I could copy/paste the new log. |
Sure, post your log output. |
18:54:24-292864 INFO Kohya_ss GUI version: v24.1.4
|
Can you look in the folder at |
Surprisingly it did. 🎉 Now if I only knew what value was messing it up. |
It's probably the |
I tried installing the newest version of Kohya, but when I try to train something, I receive the following error message. I've not idea if it's something on my end, or if it's a file that needs to be fixed.
The text was updated successfully, but these errors were encountered: