* Dump the expected and actual values for NgraphTestCase

* Adapt to changes in master

* Some docs and API unification

* Extend lp-norm functions to take bias.

* Move lp-norm utilities to nGraph core op/util.

* Move norm files to builder directory.

* Normalize fused operator implementation.

* Fused op boilerplate.

* Fix node validation and normalization across spatial axes.

* Add UT normalize across CHW with scalar scale.

* Fix expanding input tensor to 4D.

* Add more UT for 3D and 2D.

* Add more UT, with scale and across HW.

* Update to new localization of l2_norm function.

* Add type_prop UT and update gpu/igpu manifests.

* Apply clang-format.

* Add positive UT for type_prop.

* Update unit test manifests.

* Address review comments.

* Add using namespace std.

* Remove unnecessary std prefixes.

* Remove blacklisted unittests for GPU.

* Apply clang-format.

* Review comments.

* Fix clang errors.

* Basic mean normalization

* Add NVM to serializer

* Add test for mean normalization

* Add support for across_channel atribute

* Add test for mvn_mean_normalization splited by channels

* Assume that data have n and c channels

* Add support for normalize_variance attribute

* Add test for full mean variance normalization

* Add type prop test

* Skip tests on GPU

* Use ngraph builder functions instead of my own

* Update mvn.cpp

* Change order in initializer list

* Review fix

* remove throw from dtor (#2854)

* Migrate from master (#2858)

* Migrate from master. (#2866)

* Add build option NGRAPH_ENABLE_CPU_CONV_AUTO to enable/disable mkldnn convolution_auto for CPU backend. (#2868)

* Fix a bug in mkldnn emitter. (#2878)

* Fix data type for conv builder (#2885)

* Fix data type for conv builder

* Added support for other data types

* Migrate #2889 from master (#2891)

* Migrate from #2884 (#2897)

* Migrate #2864 (#2901)

* Migrate #2864

* Remove extra changes

* Copy PR #2894 to r0.19 branch (#2903)

* Extend lp-norm functions to take bias.

* Move lp-norm utilities to nGraph core op/util.

* Move norm files to builder directory.

* Apply clang-format.

* Address review comments.

* Add using namespace std.

* Review comments.

* Fix clang errors.

* Enable non-constructor use of shape inference

* Move GPU BatchNormTrainingWithStats shape inference out of the constructor

* Addressed PR feedback

* Add more slicing attributes to dynslice

* Added output shape computation for dyn slice

* Bug fixes and added unit tests

* Style fix

* Addressed PR feedback

* Move HardSigmoid to nGraph fused operators.

* UT for HardSigmoid fused operator.

* Add type_prop UT.

* Activation function parameters and hardsigmoid activation function.

* UT for lstm with hardsigmoid activation function.

* Reorder operations in implementation.

* Fix unit tests.

* Fix typo.

* Change stored activation function to pure function pointer.

* Apply style-check.

* [ONNX] Refactor LSTM tests to use NgraphTestCase

* Enable passing instance values to comparator

* Style apply.

* Fix style, syntax

* Change order of class member to fix errors.

* Switch to single-precision parameters.

* Disable unit test for IGPU.

* Fused Clamp op implementation

* Basic clamp test with some edge cases

* Dump the expected and actual values for NgraphTestCase

* Validate the min and max params for Clamp

* Use clamp in clip

* Disable Clamp and its test on iGPU

* Use getters for Clamp's parameters

* Validate clamp's params in pre_validate_and_infer_types

* Unit tests for clamp op validation

* Revert "Dump the expected and actual values for NgraphTestCase"

This reverts commit 3a029d70e62339ee84aadf2bf16e418281b85ff7.

* Clamp op docs

* Add create_dynamic_tensor function to Backend signature

* Add skeleton of dynamic wrapper class

* Simple test for construction

* Some progress on dynamic tensor creation

* Compile dynamic wrapper into core; change config string to a factory option

* Add wrapped executable, working on simple example

* Add wrapped dynamic tensor

* Update to match new tensor API

* Update wrapper to clone the graph; basic dyn test now passes

* Update call_with_validate to grok dynamic

* Make the unit tests a little prettier

* Make get_element_type virtual

* Change dynamic tests to be backend-parametric; fix get_element_type check for dyn tensors

* Add dynamic.in.cpp

* CODEOWNERS entry for dynamic_wrapper

* Remove overly-accommodating behavior for create_dynamic_tensor on non-dynamic BEs

* Pass enable_performance_collection to wrapped compile() calls

* More comments in the 'abc' test

* Fix 'create' wrapper in pybind11

* Remove unnecessary WrappedStaticTensor class

* Better names for the classes, and add some docstrings

* Fix pybind for 'create' to return shared_ptr now

* Add interpret_as_scalar helper

* Add interpret_as_scalar to OneHot

* clang-format

* Remove interpret_as_scalar from OneHot

We currently only support OneHot with constant depth

* Review comments

* Move HardSigmoid to nGraph fused operators.

* UT for HardSigmoid fused operator.

* Add type_prop UT.

* Reorder operations in implementation.

* Fix unit tests.

* Fix typo.

* Apply style-check.

* Switch to single-precision parameters.

* Disable unit test for IGPU.

* Some hacks for graph visualization

* More betterer

* Remove debug prints

* Cheap temporary hack to deal with zombie users issue

* Less hacky fix for the zombie users issue

* Change to use get_name instead of get_friendly_name; misc. cleanup of dot output

* Allow '.dot' as VisualizeTree output format, and make it default

* Cleanup, and some documentation of this jump distance nonsense

* Comment edits

* Another comment edit

* Fix typo

* Remove reverse_topological_sort function (it's not needed)

* Rewrite the MPI_Finalize

* Change destructor to virtual finalize

* Test

* More bug fix

* Delete space

* Add variable to control initialization and finalization

* Add a bool variable to control initialization and finalization for openmpi and mlsl

* wip

* Add support for storing constant array of constant values as a single values that is automatically broadcast on deserialize

* revert some changes to serializer.cpp

* fix all_close_f to support nan and inf to allow for unit test

* update unit tests to pass for all_close_f update

* fix bug with i64

* address compile issues?

* change function name to be more accurate

* fix compiler error

* [ONNX] Hardmax implementation in the onnx importer

* [ONNX] More generic handling of types in hardmax

* [ONNX] Support for doubles in EmbeddingLookup CPU builder

* [ONNX] Throw when the provided axis is out of range

* [ONNX] Skip the hardmax test on GPU

* Unused headers clean-up

* refactor: move the identity matrix generator to common.hpp

* ASSERT_VALID_ARGUMENT for axis range validation

* Adapt to the code changes in master

* [ONNX] Refactor LSTM tests to use NgraphTestCase

* Enable passing instance values to comparator

* Review comments

* Update section on Homomorphic Encryption transformer

* Make sure other updated files get added to git commit

* Clarify that rc versions may have newer features and may be less stable.

* Prettier formatting because vectors are important

* Minor review changes

* IntelGPU backend: change code owner

* Add Sevin

* quantization cleanup

* offset changed to zero point

* fix failing tests

* code style

* code style

* Use pragma clang only if compiler is Clang.

* Revert "Use pragma clang only if compiler is Clang."

This reverts commit f3deae9b59e950c9a282f9dac59e6c1a89c7c0f8.

* replace pragma clang with pragma GCC

* Add Tile op

* Added Tile op header to GPU backend

* Add dummy GPU tile emitter

* Interpreter working

* cleanup

* cleanup

* add CPU backend

* Update the rest of the backends

* update python API

* update test_case to use new shared_ptr

* Move transpose and flatten to ngraph op utils dir

* Move gemm operator into ngraph fused ops

* Style fix

* Add Gemm to serializer

* Add type_prop test for gemm

* Use Gemm default values

* Add UT for Gemm

* Fix comments

* Little cleanup

* Remove artifact headers

* Fix gemm documentation

* Skip gemm test on GPU

* Add test for broadcasting input C

* Review fix pt. 1

* Fix typo

* add dynamic function support to GraphRewrite.

* add pass property to constant folding pass.

* namespace clean up.

* fixed comment.

* Add build option NGRAPH_ENABLE_CPU_CONV_AUTO to enable/disable mkldnn convolution_auto for CPU backend.

* Add to target_compile_definitions.

* Fix style error.

* Fix the cpu thread_safe_calls test on Windows.

* Use notify_one.

* ngraph core changed

* cpu bits + pattern

* change perms

* revert std::cout to NGRAPH_DEBUG

* fix comp errors

* fix comment

* more comment fixes

* Remove callback argument

* clean up.

* fixed tests.

* more fixes for backends.

* fixe namespace.

* similar fix for recurrent matcher, and improves.

* fix build.

* Fix convbias and 1dconv for CODEGEN.

* Address PR feedback.

* Address PR feedback.

* - Adding DetectionOutput and Interpolate layers

* Adding Unit tests

* Name fix

mklml on linux has no soname set. This will cause linking against the full path for mklml when using prebuilt mkldnn and mklml. (#2851)

* Klocwork issues.

* Revert to dynamic_pointer_cast