Otherwise they are initialized concurrently in scope of MKLDNNNode
constructor without thread safety

* Firmware update with a temporary fix for conformance tests, so they wouldn't hang.

* Moved CC macros for transformations to public API (to reuse their in plugin transformations)

* Added CC to template plugin

* Moved itt.h from nGraph to openvino/cc

* Fixed build

* Fixed templatePlugin tests on IA32

* Disabled tests on MYRIAD

… dlrm (#5931)

* DEPRECATED dlrm convert instructions, corrected link in one billion model

* added DEPRECATED to dlrm title

* LRN spec refactored against explicit type indication.

* Fix spelling.

* Upload python speech sample draft

* Add function doc strings

* Add the ability to save results to a file

* Add a plugin configuration for GNA to get better accuracy

* Add errors calculation (comparison with a reference)

* Fix flake8 issues

* Add ability to run in hetero mode

* Add ability to load and save numpy format files (.npz)

* Add an error for wrong file extensions & update help message

* Add import and export GNA model options

* Add -we option to export embedded gna model

* Add readme

* Add -oname command line option (Layer names for output blobs)

* Add -iname command line option (Layer names for input blobs)

* Add info about -iname option to README.md

* doc: update readme, fix style

* Add a state reset between inferences

* add reset API to speech README

* doc: remove extra output from README

* remove onnx and TODO, format output

* Add an else branch to the if statement that checks a utterance data type

* Add dummy data for inference if a number of vectors < batch size

* Split the sample into separte files

Co-authored-by: Kate Generalova <[email protected]>

* [LPT] SplitTransformation WA for unsupported concat

* [LPT][TESTS] added test-case with Split and unsupported Concat

* CTCGreedyDecoder spec refactored against explicit type indication.

* Add backticks to output tensor description.

…lit fix (#5895)

* Azure CI: Add "Build samples" check

* Fix dir

* [LPT] ConvolutionBackpropData: handled incorrect dequantization on weights

* [LPT][TESTS] ConvolutionBackpropData: added test-cases with incorrect dequantization on weights

* Revert "Fixed nGraph lib files for Windows (#5904)"

This reverts commit 442bb4e.

* Added inference_engine_transformations.lib back on Windows

* NGraph - FrontEndManager - common API
Added Mock frontend for unit testing

* Comment out installation rules

* Fixed review comment
Introduced environment variable OV_FRONTEND_PATH for frontend paths

* Additional cmake cleanup

* More fixes of cmakelists

* Changing template file name for frontends

* Changed copyrights

* Reuse ngraph-utils for file-related operations

* Correct of frontend file extension for macos
More code coverage

* Renamed all methods to according to ngraph-style
In case of unset or empty OV_FRONTENT_PATH - load plugins from current working directory

* Fix review comments
Add clang format target for frontend_manager and mock frontend plugin

* Update docs for FrontEndCapabilities

* Use constants for flags instead of 'enum'

* clang style fix

* Fix comment: using namespace ngraph...

* Fix comment: use replace "frontend_manager" with ${TARGET_NAME} in CMakeLists.txt

* Comment fix: rename 'generic' folder to 'frontend_manager'

* Update documentation comments

* Set FrontEndManager class as 'final'

* Apply review comments

* renamed get_place_by_name_and... to get_place_by_operation_name_and...

* Separated frontend_manager.hpp into 4 header files

* Interpolate  spec refactored against explicit type indication.

* Chnaged types names.

#5914)

* Fix 'operands of different size in bitwise operation' klockwork issue.

* Changed 'byte_idx' type to size_t.

* Add extractor for ONNX GreaterOrEqual operation

* Add more extractors

* Fix extractor

Co-authored-by: achetver <[email protected]>

* Fix FakeQuantize weights fusion when multiple operations were using the same const

* [GNA] Fix scale factor calculation for Relu if it has source and doesn't have destination statistics

Co-authored-by: Dmitrii Khurtin <[email protected]>

Signed-off-by: Shoujiang Ma <[email protected]>

Co-authored-by: Gleb Kazantaev <[email protected]>

* Updated developer package generation

* Try to fix

* Reuse find_package(InferenceEngine) inside DeveloperPackage

* Fixes for ARM

* Fixed double InferenceEngineDeveloperPackage include

* WA for OpenCV

* enable make clean to remove ie_wheel artifacts

* ./setup.py:132:1: E302 expected 2 blank lines

* fix CI build issue

* set ouput dir for whl file to CMAKE_BINARY_DIR

* cleanup

* blank line

* Revise spec for nonzero op

* Applying suggestions from review comments

* Change naming in detailed description, using tuple instead of column.

* add missing newlines in attributes section

* Remove trailing space from line 7

* Fixed Roll output shape bug in RollWithEmptyAxesReplacer transformations.

* Added check in unit test.

…5883)

* Make timetests dependent on IE gflags 3rdparty from samples/

* Make stress tests dependent on IE gflags 3rdparty from samples/

* Implement Einsum reference in nGraph interpreter

Signed-off-by: Roman Kazantsev <[email protected]>

* Apply code-style patch and fix build issues

Signed-off-by: Roman Kazantsev <[email protected]>

* Fix CI build

Signed-off-by: Roman Kazantsev <[email protected]>

* Fix fast build and apply code review feedback

Signed-off-by: Roman Kazantsev <[email protected]>

* Add visitor API tests, check for input type and reduce memory consumption

Signed-off-by: Roman Kazantsev <[email protected]>

* Apply code style

Signed-off-by: Roman Kazantsev <[email protected]>

* Add visitor API tests

Signed-off-by: Roman Kazantsev <[email protected]>

* Add transformations to optimize SR model

* Add test for SplitSqueezeConcatFusion

* Add TransposeFuse tests

* Return TransposeOptimization renamed to TransposeToReshape

* Fix docstring

* Fix codestyle

* Fix build

* Fix GNA build

* Fix TransposeToReshape tests

* Fix test

* Temporarily disable cpu test

* Fix codestyle

* Fix test

* Fix test

* Enable SplitSqueezeConcatFusion

* Apply suggestions from code review

Co-authored-by: Gleb Kazantaev <[email protected]>

* Apply review feedback

* Apply review feedback

* Update split_squeeze_concat_fusion.hpp

Co-authored-by: Gleb Kazantaev <[email protected]>

…p will add unexpected padding while NMS cannot handle it properly (#6015)

* Add code style fixes

* Revert "Add code style fixes"

This reverts commit 490934f.

* Fix Invalid escape sequence

* Fix Invalid escape sequence

* Added extractor for ONNX operation Size

* Moved transformation of Size operation from TF specific to generic front phase

* Updated list of supported ONNX operation

* Moved unit test for Size decomposition to a new location

…6033)

* fix marking of nodes with shape values: corrected 'leaky' starting condition

* cleaned up code

* fixed t2t type collision error: removed strict assert from Pow; now type_infer is same as for other eltwise ops; added warnings about type alignment

* type cast for Pow according to ONNX spec

* fix highlight in dump_graph_for_graphviz

* soft_get('op')

* Revert "soft_get('op')"

This reverts commit cadfe18

* Revert "type cast for Pow according to ONNX spec"

This reverts commit bf85ebf

* applied review comments

* comment for ShapeOf bfs, removed Error from elementwise op

* reverted back dump_graph_for_graphviz: fill_color -> node_attrs

* applied last review comments: returned back Error, added type_infer into unit-test utils

* [GNA] Fix static code analysis issue 22401

* [GNA] Fix static code analysis issue 22404

* [GNA] Fix static code analysis issue 22297

* [GNA] Fix static code analysis issue 22525

* [GNA] Replace return with assert

Co-authored-by: Gleb Kazantaev <[email protected]>