[LoRA] refactor lora loading at the model-level #11719
Merged
+107
−140
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
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. |
BenjaminBossan
approved these changes
Jun 16, 2025
DN6
reviewed
Jun 18, 2025
src/diffusers/utils/peft_utils.py
Outdated
|
|
||
|
|
||
| @contextlib.contextmanager | ||
| def _lora_loading_context(_pipeline): |
There was a problem hiding this comment.
Makes sense as a context manager, but I don't see it being applied anywhere?
There was a problem hiding this comment.
@BenjaminBossan and I decided to remove its use because we preferred more explicit code to handle offloading.
DN6
approved these changes
Jun 19, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Currently, there's a significant overlap between these two methods:
diffusers/src/diffusers/loaders/peft.py
Line 121 in 8adc600
diffusers/src/diffusers/loaders/lora_base.py
Line 322 in 8adc600
This makes it challenging to incorporate new changes as the developer has to remember to propagate the changes in both locations. Hence, this PR factors out the common things from both of the places and turns them into utility functions. IMO, it also helps to make both the loading functions a bit cleaner to read.