[Suggestion] Few QoL improvements

[x02Sylvie]

1. Ability to pick huggingface model url (like swapping 1.5 stable diffusion to 1.4 or to waifu diffusion):

• https://huggingface.co/hakurei/waifu-diffusion instead of https://huggingface.co/runwayml/stable-diffusion-v1-5 for example

2. Ability to set amount of preservation/class images generated when used with prior preservation and you didn't upload your own pictures so class images are generated by colab

• I think it's related to --num_class_images but unsure

3. Training Subject: Custom/Advanced setting (helpful with waifu diffusion where you don't want 'photo of' part)

if Training_Subject=="Custom":
  PT=INSTANCE_NAME+" "+SUBJECT_TYPE
  CPT=SUBJECT_TYPE

• Thanks

[TheLastBen]

1- I'll look into it 2- the option already exists : "Number_of_subject_images" in the "Setting up" cell (default : 200) 3-I will add that category + captioned images with filenames to improve the training.