is it ok to ask question here about errors in a google colab? or is there a more appropriate chanel?
is it ok to ask question here about errors in a google colab? or is there a more appropriate chanel?
2I'm getting an error when trying to train the model, unrecognized arguments: portraits
Clooks almost spot on. as you say the eyes are the tricky one but looks good
Zthank you
CPromise last question. Once you remove backgrounds do you blip your new dataset?
Zit wouldn't let me
Zlike it kept throwing errors, i don'tthink it likes it, but you shouldn't have to caption cause your subject is the only thing in the picture
Cthank you going to now give it a go with new data set
ZGreat! can't wait to see the results. I posted some scary stuff yesterday
EFor past two days I was working with Lora and I could do training without problem. All of the sudden I can't train anymore because of the CUDA out of memory...
Is there any way to lower the vram usage for dreambooth / lora training?
Is there any way to lower the vram usage for dreambooth / lora training?
EI have 8gb vram it should work with lora 
Zdoes anyone know a good way of converting prompts between systems such as auto1111, invoke ai, midjourney?
Zwonder what's using your vram
Fyou can ask here
Fplease show error message
Fyes it should work. must be a settings error
2train_dreambooth.py: error: unrecognized arguments: portaits
EI have comeback to my old settings and it's working. I have to check what is different
2it appears there is an argument in the command without -- in front of it
Fplease show screenshot
2
2subprocess.CalledProcessError: Command '['/usr/bin/python3', 'train_dreambooth.py', '--pretrained_model_name_or_path=runwayml/stable-diffusion-v1-5', '--pretrained_vae_name_or_path=stabilityai/sd-vae-ft-mse', '--output_dir=/content/drive/MyDrive/stable_diffusion_weights/Images', 'portaits', '--revision=fp16', '--with_prior_preservation', '--prior_loss_weight=1.0', '--seed=1337', '--skip_portraits', '--resolution=512', '--train_batch_size=1', '--train_text_encoder', '--mixed_precision=fp16', '--use_8bit_adam', '--gradient_accumulation_steps=1', '--learning_rate=1e-6', '--lr_scheduler=constant', '--lr_warmup_steps=840', '--num_class_images=1260', '--sample_batch_size=4', '--max_train_steps=8400', '--save_interval=10000', '--save_sample_prompt=emsharan', '--concepts_list=concepts_list.json']' returned non-zero exit status 2.
2this is the final line of text that is too long for the screenshot
2I put the part that i think is problematic in bold
2the '--skip_portraits' is something i tried but did not work, i removed that again
2I'm currently trying to go over all the steps again in a new worksheet
Fok let me check
Fyou are following this video right
FYouTubeSECourses - Software Engineering Courses
Our Discord : https://discord.gg/HbqgGaZVmr. How to do free Stable Diffusion DreamBooth training on Google Colab for free. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on 
https://www.patreon.com/SECourses
Playlist of Stable Diffusion Tutorials, Automatic1111 and Goo...
https://www.patreon.com/SECoursesPlaylist of Stable Diffusion Tutorials, Automatic1111 and Goo...
2yes
Fyou need to use this google colab
Fthere is no such command as --skip_portraits
Fwhere do you get that?
2yes, I found that out, was a wild guess based on other possible skip command, did not know what to do
Fthe only command you can add additionally is
F--gradient_checkpointing
Fthat reduces vram usage
2is this a new colab?
Fif you get out of memory error you can add that
Fthis colab constantly gets updated
Fwith changes
Fthis is the one in the description
2I'll try this out, thank you very much
Fthis is the one i linked in description you didnt see it?
2hmm i used the one in the description
Fyes this is that one
2ok, thx, i'll try again
EThis graph from dreambooth indicating losses is reliable?
Sare you guts changing the learning rates in textural inversion or using constant learning rate?
Nhi there. I am asking myself (and you here) what would be the best way to chieve the best results when I have a trained image like this (https://huggingface.co/plasmo/woolitize-768sd1-5) and want to ake a woolitized picture of myself.
I trained that image on my face, but it seems to loose its "woolitizing skills". what could i try instead?
I trained that image on my face, but it seems to loose its "woolitizing skills". what could i try instead?
