Skip to content
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.

Sign up for GitHub

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

Jump to bottom

Add initial support for full SDXL finetuning #124

Merged
merged 33 commits into from
May 13, 2024
Merged

Add initial support for full SDXL finetuning #124

merged 33 commits into from
May 13, 2024

Conversation

RyanJDick
Copy link
Collaborator

  • Add new SDXL_FINETUNE pipeline. (Not added to the GUI yet)
  • Add script for LoRA extraction from a finetuned UNet model.
  • Add a tutorial with instructions for using the new pipeline.

RyanJDick added 30 commits May 2, 2024 15:39
@RyanJDick
Fix default model paths for SDXL pipelines.
6b25406
@RyanJDick
Initial config file for the SDXL_FINETUNE pipeline.
4fc326d
@RyanJDick
Copy SDXL_LORA train.py as a starting point for the SDXL_FINETUNE pip…
bc4c49d 
…eline.
@RyanJDick
First version of SDXL finetune pipeline. OOMs with 24GB.
7d1bd63
@RyanJDick
Add support for bitsandbytes, and 8-bit Adam.
9a00a9d
@RyanJDick
Minor fix to get SDXL finetune script working.
36a06bb
@RyanJDick
SDXL finetune config tuning.
b600059
@RyanJDick
More tuning of the SDXL finetune sample config.
f858f13
@RyanJDick
Fix tensorboard prefix for the SDXL finetune pipeline.
48c5e74
@RyanJDick
Add checkpoint saving to the SDXL finetune pipeline.
173584a
@RyanJDick
Copy the LoRA extraction script from https://raw.githubusercontent.co…
1ed40b3 
…m/kohya-ss/sd-scripts/bfb352bc433326a77aca3124248331eb60c49e8c/networks/extract_lora_from_models.py as a starting point.
@RyanJDick
Add source attribution to the LoRA extraction script.
2d31981
@RyanJDick
Tidy logger and str_to_dtype() in lora extraction script.
f04d5ec
@RyanJDick
Simplify model type selection in lora extraction script.
fb41d98
@RyanJDick
Tidy model loading in LoRA extraction script.
640c155
@RyanJDick
Get diff calculation in the LoRA extraction script working.
42381c6
@RyanJDick
Add torch.no_grad() decorator to the entire lora extraction script.
df5b10d
@RyanJDick
Move core SVD logic to a separate function in the lora extraction scr…
734906c 
…ipt.
@RyanJDick
Tidy extract_lora_from_diffs() and fix linter errors.
d6f44fb
@RyanJDick
Fix LoRA saving logic in LoRA extraction script. Satisfy linter.
365ebbb
@RyanJDick
Get the LoRA extraction script working for the first time.
0602baf
@RyanJDick
Add SDXL finetune config to the reference docs.
3bce40e
@RyanJDick
Add save_checkpoint_format config to the SDXL finetune pipeline.
4f11aef
@RyanJDick
Add save_dtype to SDXL finetune pipeline.
21ae581
@RyanJDick
Minor improvements to LoRA extraction.
24524e7
@RyanJDick
Run lora extraction on calculations on CUDA device.
caddfd9
@RyanJDick
Remove empty placeholder documentation files.
82d1cad
@RyanJDick
Add robocat images to docs.
8d3ee8e
@RyanJDick
Compress the bruce_the_gnome images in the docs - they were huge.
a75ad73
@RyanJDick
(minor) Fix typo in docs.
072b78c
@RyanJDick RyanJDick merged commit 953b54a into main May 13, 2024
1 check passed
@RyanJDick RyanJDick deleted the full-sdxl-finetune branch May 13, 2024 21:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@RyanJDick