Fix custom ops loading in diffusers #1655
Open
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
Signed-off-by: Daniel Socek <[email protected]>
libinta
reviewed
Dec 20, 2024
| @@ -165,8 +165,6 @@ def __init__( | |||
| "`torch_dtype=torch.bfloat16` was given. Disabling mixed precision and continuing in bf16 only." | |||
| ) | |||
| self.gaudi_config.use_torch_autocast = False | |||
| else: | |||
| self.gaudi_config.declare_autocast_bf16_fp32_ops() | |||
|
|
|||
There was a problem hiding this comment.
where did you set this after change?
|
@regisss Yes, here's how: #1657 is a temporary, less intrusive, and partial fix that can be implemented with minimal validation and testing, ensuring it doesn’t delay the release. This PR, on the other hand, is the complete fix. However, @libinta recommended postponing it until after the release, as it requires extensive validation. |
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Outdated suggestions cannot be applied.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 does this PR do?
This PR has critical fix for custom ops loading in diffusers.
More information
As discussed in PR #1631, removing
htcoreimport before model loading would break quantization support for OH diffusers; however, @skaulintel encountered issues with some workloads whenhtcoreis imported before model is loaded. The underlying issue happens to be due to how GaudiConfig is handled when custom ops precision lists are defined in the configuration (e.g. in Habana/stable-diffusion-2).This PR fixes the underlying issue.