experiment(backend): autocast dtype in CustomLinear #7843
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
This resolves an issue where specifying `float32` precision causes FLUX Fill to error. I noticed that our other customized torch modules do some dtype casting themselves, so maybe this is a fine place to do this? Maybe this could break things... See #7836
psychedelicious
requested review from
lstein,
blessedcoolant,
brandonrising,
hipsterusername and
jazzhaiku
as code owners
github-actions
bot
added
python
invokeai/backend/model_manager/load/model_cache/torch_module_autocast/cast_to_dtype.py
Outdated
Show resolved
Hide resolved
| @@ -73,6 +74,10 @@ def _autocast_forward_with_patches(self, input: torch.Tensor) -> torch.Tensor: | |||
| def _autocast_forward(self, input: torch.Tensor) -> torch.Tensor: | |||
| weight = cast_to_device(self.weight, input.device) | |||
| bias = cast_to_device(self.bias, input.device) | |||
|
|
|||
| weight = cast_to_dtype(weight, input.dtype) | |||
There was a problem hiding this comment.
This is probably fine, but some models may act weirdly due to potential precision loss if we provide inputs with less precision than the model 🤔 In an ideal world I'd think we'd want to ensure the precision of the inputs are compatible with the model before calling it
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Outdated suggestions cannot be applied.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.
Summary
This resolves an issue where specifying
float32precision causes FLUX Fill to error.I noticed that our other customized torch modules do some dtype casting themselves, so maybe this is a fine place to do this? Maybe this could break things...
See #7836
Related Issues / Discussions
Closes #7836
QA Instructions
Try various model combos. I don't know what I'm doing and this could be a Bad Idea™️.
To reproduce the problem in the linked issue, set
precision: float32ininvokeai.yaml, then try to use FLUX Fill.Merge Plan
n/a
Checklist
What's Newcopy (if doing a release after this PR)