Record the model_variant in t2i and clip_vision configs #5989
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
- Move base of t2i and clip_vision config models to DiffusersBase, which contains a field to record the model variant (e.g. "fp16") - This restore the ability to load fp16 t2i and clip_vision models - Also add defensive coding to load the vanilla model when the fp16 model has been replaced (or more likely, user's preferences changed since installation)
lstein
requested review from
blessedcoolant,
GreggHelt2,
brandonrising,
RyanJDick and
hipsterusername
as code owners
github-actions
bot
added
python
hipsterusername
approved these changes
Mar 19, 2024
psychedelicious
approved these changes
Mar 19, 2024
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Outdated suggestions cannot be applied.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.
What type of PR is this? (check all applicable)
Have you discussed this change with the InvokeAI team?
Have you updated all relevant documentation?
Description
Diffuser models downloaded from huggingface have repo variants, including the commonly-used
fp16variant. Both T2I and clip_vision models can havefp16variants, but at some point during the model manager refactor the base for their pydantic config models was moved fromDiffusersConfigBaseto a simpler base. Hencefp16models would not load. This PR corrects the problem.It also adds defensive code to the diffusers loading method. If a repo variant is requested but not available, the loader will attempt to load the default version. This will have the effect of protecting against a family of potential edge cases in which the user has renamed a model, cloned a huggingface repo, overwritten a weights file, or is going back and forth between different versions of InvokeAI while using the same models directory.
Related Tickets & Documents
QA Instructions, Screenshots, Recordings
TencentARC/t2i-adapter-lineart-sdxl-1.0models/sdxl/t2i_adapter/lineart-sdxl.0/diffusion_pytorch_model.fp16.safetensorsMerge Plan
Merge when approved.
Added/updated tests?
[optional] Are there any post deployment tasks we need to perform?