-
Notifications
You must be signed in to change notification settings - Fork 2.4k
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
[ONNX] Added initial support of DequantizeLinear-21 #26915
Conversation
95d7a72
to
05f49a2
Compare
Removed commented code
src_x.get_shape()[0] % block_size == 0, | ||
"DequantizeLinear doesn't support case when first dimension of X cannot be divided by block_size"); | ||
|
||
const auto& x = src_x.get_element_type() == scale_type ? src_x : std::make_shared<v1::ConvertLike>(src_x, scale); |
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.
can you please simplify? head breaking expression:)
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.
Ok, in a separate PR, CI is finally green
FRONT_END_GENERAL_CHECK(scale_shape.rank().is_static(), "Rank of the input data tensor has to be known (static)."); | ||
FRONT_END_GENERAL_CHECK(scale_shape.rank().get_length() == 2, | ||
"DequantizeLinear cannot operate with more than 2D scales"); | ||
FRONT_END_GENERAL_CHECK(src_x.get_partial_shape().is_static(), |
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.
really important limitation that can be often not met. We can think of how to support it via decomposition one more time, or request new op in the opset.
Details:
Tickets: