Skip to content
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.

Sign up for GitHub

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

[LPT] ReduceSum: zero point support #24977

Merged
merged 2 commits into from
Jun 13, 2024
Merged

[LPT] ReduceSum: zero point support #24977

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
48526d8
[LPT] ReduceSum: zero point support
eshoguli Jun 12, 2024
4f44307
comments
eshoguli Jun 12, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 4 additions & 1 deletion src/common/low_precision_transformations/src/reduce_sum.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,7 +68,10 @@ void ReduceSumTransformation::changeDequantizationValues(
}

// (a1 - s) + (a2 - s) + ... + (an - s) = (a1 + a2 + ... + an) - n * s
const auto reductionSizeConstant = ov::opset1::Constant::create(deqPrecision, Shape{}, { static_cast<float>(reductionSize) });
const auto reductionSizeConstant = ov::opset1::Constant::create(
dequantization.subtractConstant->get_element_type(),
eshoguli marked this conversation as resolved.
Show resolved Hide resolved
Shape{},
{ static_cast<float>(reductionSize) });
const auto result = fold<ov::opset1::Multiply>(dequantization.subtractConstant, reductionSizeConstant);

replace_node(dequantization.subtractConstant, result);
Expand Down
65 changes: 65 additions & 0 deletions src/common/low_precision_transformations/tests/reduce_sum_transformation.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ using namespace ov::builder::subgraph;
class ReduceSumTransformation : public ReduceTransformation<ov::op::v1::ReduceSum> {
void SetUp() override {
ReduceTransformation::SetUp();

const auto transformationParams = std::get<1>(GetParam()).params;

SimpleLowPrecisionTransformer transform;
Expand Down Expand Up @@ -336,4 +337,68 @@ INSTANTIATE_TEST_SUITE_P(
::testing::ValuesIn(reduceSumTransformationTestValues)),
ReduceSumTransformation::getTestCaseName);
} // namespace testValues2

namespace testValues3 {
const std::vector<ov::PartialShape> inputShapes = {
{4, 3, 16, 16}
};

const std::vector<ReduceTransformationTestValues> reduceSumTransformationTestValues = {
{
LayerTransformation::createParamsU8I8(),
{0},
true,
{
ov::element::u8,
{
{ov::element::f32},
{{2.f, 4.f, 6.f}, ov::element::f32, {1, 3, 1, 1}, false, 1ul, ov::element::u8, true},
eshoguli marked this conversation as resolved.
Show resolved Hide resolved
{{0.1f, 1.f, 10.f}, ov::element::f32, {1, 3, 1, 1}}
}
},
{
ov::element::u8,
{},
ov::element::f32,
{
{},
{{8.f, 16.f, 24.f}, ov::element::f32, {1, 3, 1, 1}, false, 1ul, ov::element::u8, true},
{{0.1f, 1.f, 10.f}, ov::element::f32, {1, 3, 1, 1}}
}
}
},
{
LayerTransformation::createParamsU8I8(),
{0},
true,
{
ov::element::i8,
{
{ov::element::f32},
{{2.f, 4.f, 6.f}, ov::element::f32, {1, 3, 1, 1}, false, 1ul, ov::element::i8, true},
{{0.1f, 1.f, 10.f}, ov::element::f32, {1, 3, 1, 1}}
}
},
{
ov::element::i8,
{},
ov::element::f32,
{
{},
{{8.f, 16.f, 24.f}, ov::element::f32, {1, 3, 1, 1}, false, 1ul, ov::element::i8, true},
{{0.1f, 1.f, 10.f}, ov::element::f32, {1, 3, 1, 1}}
}
}
}
};

INSTANTIATE_TEST_SUITE_P(
smoke_LPT,
ReduceSumTransformation,
::testing::Combine(
::testing::ValuesIn(inputShapes),
::testing::ValuesIn(reduceSumTransformationTestValues)),
ReduceSumTransformation::getTestCaseName);
} // namespace testValues3

} // namespace
2 changes: 1 addition & 1 deletion src/core/src/op/multiply.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,7 @@ bool Multiply::evaluate(TensorVector& outputs, const TensorVector& inputs) const
this,
outputs,
inputs,
OV_PP_ET_LIST(f32, f64, i32, i64, u32, u64),
OV_PP_ET_LIST(f32, f64, i8, i32, i64, u8, u32, u64),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why do we need these changes?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

to support i8 & u8 reference

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I meant why do we need to support i8/u8 reference in this PR ;) It seems like we don't need these changes. However, I was just curious, don't see anything bad in that. We can leave it as is

multiply::Evaluate,
inputs[0].get_element_type(),
inputs[0],
Expand Down
Loading