-
Notifications
You must be signed in to change notification settings - Fork 255
Aanuf/sdpa v fp8 #3485
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: develop
Are you sure you want to change the base?
Aanuf/sdpa v fp8 #3485
Conversation
@@ -333,6 +333,7 @@ def __init__( | |||
post_processing_marker_metatypes: Optional[list[type[OperatorMetatype]]] = None, | |||
metatypes_to_ignore: Optional[list[type[OperatorMetatype]]] = None, | |||
scales_unification_map: Optional[dict[type[OperatorMetatype], list[type[OperatorMetatype]]]] = None, | |||
is_fp8: bool = False, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@andreyanufr, @AlexanderDokuchaev, please provide suggestion how to avoid passing is_fp8 parameter in the solver.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As we discussed offline, please add support for attention subgraph without sdpa via disabling ignored patterns. cc' @xiao1228
Changes
FakeConverr for V tensor for SDPA layer in the case of FP8 quantization for NPU performance.
Reason for changes
Related tickets
CVS-166427
Tests
In process.