bmaltais comments

Results 726 comments of


                                            bmaltais

Attribute error even thought I already changed the train_util.py document

OK... look like you don't have enough VRAM for the settings you are usiing. You only have 6GB of VRAM and this is probably not enough. 8 GB is a...

Attribute error even thought I already changed the train_util.py document

You have already pretty reasonable values... not sure what else you could try... Maybe try going to LoRA of 64 DIM instead of 128?

persistent error: no module named library

This is the case when the library is not installed as part of requirements.txt. try running upgrade.ps1 to rerun the requirements installation.

[Feature Request] Loading the json configuration without a modal window

Yeah... it is annoying me also... especially when I paste the path to the json file and it still open the dialog... I will try to implement the 4th button...

[Feature Request] Loading the json configuration without a modal window

OK, I have implemented the 1st thing to load the config without opening the modal window. It is pushed to the dev branch. Now... for the 2nd it is a...

CUDA_SETUP: WARNING! libcudart.so not found

When is it spitting this error. This is normal when using tensorboard... but nor when training.

Stuck at caching latents

The sd-script is really meant to be used with an NVidia card. I am not aware of it running properly under an AMD GPU...

train_network.py: error: argument --caption_dropout_every_n_epochs: invalid int value: '1.0'

I will convert the field to only accept integer and use that instead of a float... That you for reporting.

train_network.py: error: argument --caption_dropout_every_n_epochs: invalid int value: '1.0'

The issue is fixed in the latest dev branch commit

error when i start train model

I noticed this error. I think your windows pagefile is too small. Try increasing the page file system size: ``` The error message "OSError: [WinError 1455] Le fichier de pagination...