huggingface
[WIP] Add inpainting training script
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.
Gave this script a try it seems like it cannot be used for fine-tuning an existing sdxl inpaint model and can only be used as creating an inpainting model from base sdxl.
The error says that the number of input channels in the unet required should be 9 (base sdxl) but is 19 (inpainting sdxl). I don't know if it is possible to scale the input tensor channel from 9 to 19 since the input already contains noisy + mask + mask_image latents.
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
Hey @patil-suraj, thanks for sharing this training code.
Could you please share with us the exact training args (without dataset sharing) you used to train inpainting model including number of GPUs used (to get total batch size)? Also the dataset size would help a lot. Also would appreciate sharing wandb run if possible 🤗
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.