Model merging utility scripts #138
Merged
Conversation
This file contains 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
…n result in drastically higher memory utilization (it seems like ther might be a bug in diffusers causing multiple copies of the weights to be held in memory).
…r structures / APIs.
… solve issue with dtype being changed during the calculation.
…it easier to support more submodels.
… diffusers and checkpoint files, and to support unet, te1, and te2 submodels.
…eliminate benign error logs.
…ther model merging scripts.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
This PR adds some new model merging utility scripts (LERP, SLERP, TIES, DARE), and improves the existing LoRA merge scripts.
The current list of model merging scripts is:
extract_lora_from_model_diff.py: Extract a LoRA model that represents the difference between two base models.merge_lora_into_model.py: Merge a LoRA model into a base model to produce a new base model.merge_models.py: Merge 2 or more base models to produce a single base model (using either LERP or SLERP).merge_task_models_to_base_model.py: Merge 1 or more task-specific base models into a single starting base model (using either TIES or DARE). This merge strategy aims to preserve the task-specific behaviors of the task models while making only small changes to the original base model.Left for future work:
merge_task_models_to_base_model.py.