bmaltais
bmaltais
The python code is made to run in a venv and I don't know the extent of changes to make thios possible... I will look at it when I have...
What is the error message you are getting? THe python code set the PYTHONPATH internally as: ```python env["PYTHONPATH"] = ( fr"{scriptdir}{os.pathsep}{scriptdir}/sd-scripts{os.pathsep}{env.get('PYTHONPATH', '')}" ) ``` Perhaps I would check for the...
I have created a new branch called `portable-gui` to implement the changes... You can use it for testing... I moved the environment code into common_gui.py so you can easilly modify...
I will let others provide help on this one as I don't have experience training on a100 with large dataset... You may want to make sure you are cachiong latents...
Instead of using the meta_*.json files you can use a .toml config file. Try opening the `./test/config/finetune-AdamW-toml.json` for a working example... not sure how it will do under runpod as...
Do you mean if you installed is as part of the optional setup option in setup.bat? One way (the nucrear approach) is to delete the venv folder and the rerun...
Have you tried switching to us a dataset toml config file instead of the old folder based image finetuning? Load the test example and study how the dataset timl file...
Dans, I will fix that today.
OK... I never use this option so maybe there is something I don't get... vae are usually indifidual files... so I check for the presence of the file at the...
Crap... this is what I was afraid of... I will update the model verification and merge the PR on top.