Skip to content

Conversation

@pcuenca
Copy link
Member

@pcuenca pcuenca commented Dec 13, 2022

Bring dreambooth changes in #1668 to the rest of the training scripts.
Fixes #666.

@HuggingFaceDocBuilderDev
Copy link

HuggingFaceDocBuilderDev commented Dec 13, 2022

The documentation is not available anymore as the PR was closed or merged.

Copy link
Contributor

@patrickvonplaten patrickvonplaten left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would maybe rephrase the help message a bit and possible even add an example of how to use intermediate checkpoints for inference

@patrickvonplaten
Copy link
Contributor

BTW @pcuenca , accelerate already merged the "only last checkpoitns" features to "main": huggingface/accelerate#916

Let's maybe allow people to directly use it depending on the accelerate version?

Co-authored-by: Patrick von Platen <[email protected]>
@pcuenca
Copy link
Member Author

pcuenca commented Dec 13, 2022

BTW @pcuenca , accelerate already merged the "only last checkpoitns" features to "main": huggingface/accelerate#916

Let's maybe allow people to directly use it depending on the accelerate version?

Yes!

--max_train_steps=800
```

### Saving checkpoints while training
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Very nice!

Copy link
Contributor

@patrickvonplaten patrickvonplaten left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Perf thanks!

@pcuenca pcuenca merged commit badddee into main Dec 15, 2022
@pcuenca pcuenca deleted the resume-checkpoints branch December 15, 2022 18:49
sliard pushed a commit to sliard/diffusers that referenced this pull request Dec 21, 2022
* Add state checkpointing to other training scripts

* Fix first_epoch

* Apply suggestions from code review

Co-authored-by: Patrick von Platen <[email protected]>

* Update Dreambooth checkpoint help message.

* Dreambooth docs: checkpoints, inference from a checkpoint.

* make style

Co-authored-by: Patrick von Platen <[email protected]>
@sanghyun0927
Copy link

Sorry for that I am newbie in github

Just want you to know that this line may have to be changed

examples/research_projects/dreambooth_inpaint/train_dreambooth_inpaint.py

675 for epoch in range(first_epoch, args.num_epochs):
==>
675 for epoch in range(first_epoch, args.num_train_epochs):

@pcuenca
Copy link
Member Author

pcuenca commented Jan 2, 2023

I think you are right, @sanghyun0927, just opened #1885 to fix it. Thanks a lot!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

How to resume from checkpoint.

5 participants