* Review slice ope for
- Interval dimension and label propagation
- add template shape inference with static shape test
- check preserve partial values on inputs
- check upper/lower evaluate
* Add bounds evaluation for inputs start, stop
* Share code between slice and strided slice
Use same function to calculate elements in step
* Add array includes
* Add to int64_t strides size
* Fix windows compile warnings
* Fix shape inference for unknown axes
* Remove empty lines in slice shape inference
* Correct slice static shape tests
* Use arrays of const chars to store literals
Remove and update exception messages for strided slice
* Fix slice test and apply review comments
* Fix compilation issues
* Fix ellipsis when there is not begin
* Fix get element type for const inputs
* Insert optional axes as const or dynamic param
* Remove temp vectors for dimensions calculation
* Revert set optional input in ctor
* Fix forward slicing for negative start and MAX end
Previously, reorder_input pass checked only output layouts of each node, with an assumption that the input/output ranks are same at that pass.
However in dynamic shape using ngraph shape infer, there are cases input / output ranks differ. In such cases, the reorder_input pass inserts reorder to the format of current node's output_layout in the input of the current node, which causes error.
Fixed the above behavior by applying set_preferred_input/output_layout.
* Fixed Windows Arm64 build
* Fixed linker warnings in docs
* Fixed Windowd ARM build
* Reverted changes in docs
* Added OpenVINO level macro to define CPU architecture
* Properly remove xbyak from reference kernels optimizations
* Added documentation for cmake functions
* Fixed samples build
* Added installation dir for host protoc
* Fixed installation stage for host_protoc
* Added build dir
* Fixed compilation on Apple
* cççç
* Fixed clang code-style
* Tely on CPU id from xbyak
* Fixed build
* Fixed compilation on mac
* Fixed compilation
* [TF FE] Support Group Convolutions
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* Split cases of regular Convolution and GroupConvolution operations
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* fix checking if there are data nodes for operations with several outputs
* added unit-test
* added ports explicitly, removed redundant lambda
* typo in port
Update POT tool configs to use unified scales for concat by default.
Co-authored-by: Alexander Kozlov <alexander.kozlov@intel.com>
Co-authored-by: Liubov Talamanova <liubov.talamanova@intel.com>
It allows to fix inference results mismatch for some models
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* added missed layout to string conversion
* added an unit test for layout_from_string
* added missing spaces around =
* replaced 95 with SCALAR
* moved layout_to_string into a new file
* changed to parametrized test
* [GPU] Use in-order queue for devices with dpas support
* [GPU] Don't query properties in inference time
Co-authored-by: Mingyu Kim <mingyu.kim@intel.com>
* Add ov::label_t alias and set it to uint32_t
to simplify conversion to tensor element type
* Fix clang formatting
* Fix one hot test to use label_t
* Remove label types alias duplicates
propagate using alias for label types
* Correct label type in transpose tests
* Define label_t in dimension
remove size_t usage as label type.
* Correct label type usage in Einsum and LSTM op's
* Correct label type usage in tracker transformation
scatter update and split op tests
* Correct usage label type in type_porp tests
- squeeze
- tile
- transpose
- variadic split
* Correct usage label type in type_porp tests
- squeeze
- transpose
- unsqeeze
- variadic split
* Remove excess Azure Pipelines steps
* Installed unzip inside Azure pipelibes
* Added --no-install-recommends
* Move ccache to build tools
* Added libffi-dev for python _stypes
* Exit if debian packages cannot be installed
* Added comment
* Install wget for cmake downloading
* Update Azure
* Try to fix samples tests on Debian Linux
* Fixes
* Moved env
* More refactoring
* Fixes
* Revert: bash => script back
* Fixed arm64 linux
* Fixed CUDA
* Fixed mac
* Try to fix
* Try to fix cuda
* Fixed CUDA
* Fixes
* Optimize Mac build
* Fixed windows
* Try to fix debian
* Build CC COLLECT
* Use ninja for CC
* Try to fix WinCC
* Apply suggestions from code review
* Fixed sample spath
* Update .ci/azure/windows_conditional_compilation.yml
* Apply suggestions from code review
* Fixed CC ON compilation with ninja
* Improvements
* Print CC header content
* Try to fix Ninja build for Win CC
* Added explicit dependency between targets
* Reverted changes in Windows CC Azure config
* Update cmake/developer_package/compile_flags/os_flags.cmake
* Support oneDNN3.0 API
* Use binary_mul post_opt instead of oscale channel-wise mask(2)
* Disable some post-opt fusing because of no eltwise scale API
eltw(non_linear)+eltw(linear), eltw+sum+eltw(linear)
* Fix hardwish issue in 3.0
hard coded hardswish parameter(2.7) is changed alpha and beta from user's required input.
Signed-off-by: hyunback <hyunback.kim@intel.com>
* keras to tflite version of tests
* Update tests/layer_tests/common/tf2_layer_test_class.py
Co-authored-by: Evgenya Stepyreva <eva.my.link@gmail.com>
* moved out tf utility functions from modules with tf_layer_test classes to tf_utils module
* moved out tf utility functions from modules with tf_layer_test classes to tf_utils module and tflite_utils modules
Co-authored-by: Evgenya Stepyreva <eva.my.link@gmail.com>
* Add common test api to generate unique test file names
* Use unique file names in ir frontend tests
* Use unique file names in plugin tests
* Fix code style
* fix1
* fix2
* Add unique file names for other tests
* Fix code style 2
* fix3
* AUTO CTPUT single device always goes single device logic
* CreateInferRequest goes single device logic when perf hint is ctput
* Optimize single device logic according to bell opinion
* Add code comments
* merge two isCumulative conditions
* add ctput test case
* Remove the redundant header file of auto_ctput_test.cpp
* Modify the test case according to bell comments
