* Enable low precision types for ConvertLike operation
* Migrate backend unit test suite to template plugin reference tests
* Fix typo in naming convention
* Avoid duplication of template plugin tests execution
* Add file to instantiate TEST_P and avoid test execution duplication
* new implementation of FakeQuantize
* move FakeQuantize backend test to fake_quantize.in.cpp
* enable some test for FakeQuantize
* Add support for PDPD broadcasting and some backend tests
* arm plugin expect less attributes in function call
* try to fix arm plugin build
* try to build arm plugin
* start changing backend test
* add check for attributes number in visitor test
* fix backend test after refactoring
* add default parameter value to runtime::reference::fake_quantize
* Revert "add default parameter value to runtime::reference::fake_quantize"
This reverts commit 9d2c00d967.
* add SLT for FakeQuantize
* remove fesetround
* change `v1` namesapce to `v0` in ref impl
* add FakeQuantize-1 to VERIFIED_OP_REFERENCES
* pass immutable reference to test functions
* Written header file for the nGraph reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Small fixes.
* Started to write cpp-file for the nGraph reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Written the postprocessing function for the nGraph reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Code style fixes.
* Started to write calculations of the nGraph reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Written the helper function refine_anchors().
* Written the helper function unpack_boxes().
* Small fix.
* Code style fixes.
* Written calculations of the nGraph reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Code style fixes.
* Fixed type of one argument of fill_output_blobs().
* Code style fix.
* Written evaluate() for the operation ExperimentalDetectronGenerateProposalsSingleImage (in evaluates_map.cpp).
* Deleted redundant 'using' for typedef.
* Continued to write tests for ExperimentalDetectronGenerateProposalsSingleImage evaluation in the nGraph reference implementation.
* Fixed the reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Deleted commented code. Code style fixes.
* Code style fix.
* Deleted debug prints in the test for an evaluation of the nGraph operation ExperimentalDetectronGenerateProposalsSingleImage.
* Enabled ONNX Importer test for the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Enabled some test.
* Small fix.
* Now postprocessing of the nGraph reference realization of the operation ExperimentalDetectronGenerateProposalsSingleImage throws exception when input data type is not supported.
* Code style fix.
* Written yet another test for the nGraph reference implementation of the operation ExperimentalDetectronGenerateProposalsSingleImage.
* Reused ProposalBox type from the reference implementation of the operation Proposal.
* Added check for the attribute post_nms_count in the operation ExperimentalDetectronGenerateProposalsSingleImage evaluation.
* Small fixes.
* Add output_padding and output_shape into GroupConvolutionBackprop SLTs.
* Add legacy API for arm-plugin dependency.
* Replaced templated input, output and filter with one template.
* Apply correct format.
* Change to use INSTANTIATE_TEST_SUITE_P macro.
* Implement Einsum reference in nGraph interpreter
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Apply code-style patch and fix build issues
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Fix CI build
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Fix fast build and apply code review feedback
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Add visitor API tests, check for input type and reduce memory consumption
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Apply code style
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Add visitor API tests
Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
* Started to write nGraph reference implementation of the operation ExperimentalDetectronTopKROIs.
* Written cpp-file for the nGraph reference implementation of the operation ExperimentalDetectronTopKROIs.
* Code style fixes.
* Started to write evaluate() (in evaluates_map.cpp) for the nGraph operation ExperimentalDetectronsTopKROIs.
* Code style fixes.
* Written evaluate() for ExperimentalDetectronTopKROIs.
* Small fixes.
* Written test for ExperimentalDetectronTopKROIs evaluation.
* Small fix.
* Fixed Apache license header.
* Some changes.
* int64_t was replaced with size_t in the body of ngraph::runtime::reference::experimental_detectron_topk_rois().
* Added more test for evaluation of ExperimentalDetectronTopKROIs.
* Add Convert unit tests to check C++ to LP types convertion.
* Add unit tests to check LP types to C++ ints.
* Disable LP to C++ ints tets in CPU plugin.
* Remove f64 support from Convert evaluate.
* Added tests for 2D DFT when we have signal_size and axes [0, 2].
* Rewritten reference implementation of DFT and IDFT.
* Code style fixes.
* Reverted changes.
* Added debug prints.
* Temporarily commented some functions.
* Reverted changes in (I)DFT reference.
* Added more debug prints (in evaluates_map.cpp, to evaluate() for IDFT).
* Added setting output shape for DFT and IDFT evaluate() in evaluates_map.cpp.
* Added more tests for signal_size case of DFT.
* Written tests for signal_size case of IDFT.
* Deleted debug prints from (I)DFT reference implementation (the file fft.cpp).
* Deleted debug prints from evaluate() for DFT and IDFT.
* Deleted debug prints from tests for DFT.
* Deleted debug prints from IDFT tests.
* Code style fixes.
* Enabled CPU evaluation tests for DFT and IDFT in CPU supported cases.
* Enabled CPU (I)DFT layer tests.
* Copied fix for specialization of function is_float() for (b)float16 from Pertovsky Aleksandr's PR 5654.
* Reverted changes in the function is_float() from ie_precision.h.
* Disabled BF16 CPU FuncTests for (I)DFT.
* Deleted filling 'buffer' by zeros. Simplified the function lengths_except_given_axis.
* Small fix.
* Code style fix.
* Reverted deleting 'std::fill(buffer.begin(), buffer.end(), complex_type{0.0f, 0.0f});'
* Reverted changes in the function lengths_except_given_axis().
* Changes in the function lengths_except_given_axis().
* Small fix.
* Deleted commented code.
* Simplifications in the function lengths_except_given_axis().
* Deleted commented code.
* Enabled BF16 (I)DFT CPU FuncTests.
* Fixed comments before disabled (I)DFT IE_CPU tests.
* [IE TESTS] Fix comparation in LayerTestUtils
* Fixes
* Small fix
* Int4 fixes
* remove extra
* Fix NMS
* Some fixes for tests
* Add small fix
* [IE TESTS] Remove const folding as a result engine
* Remove extra
* Revert remove constant folding (DSR test) & fix some cases for cpu
* Fix GNA
* Reference implementation init
* Backend tests
* Single layer tests
* Update offset range in layer tests
* Align int types with ng op
* Update spatial bins type
* Type update
* Fix sub bin calculation in mkldnn plugin
* Update summarize py file
* Align result type
* Refactoring
* Apply review comments
* Add serialize layar tests
* Adjust int comparison
* Adjust code style
* Use clamp reference
* Unify style
* Additional check for negative output dim
* Set tensor output shape in evaluate
* Add visit attributes test
* Small refactor
* Code style (namespace comments)
* Fix CommonTestsUtils::fill_data_roi usage.
This function was generalized in PR #5432 and its siganutre has changed.
* Update licenese header with mention about original authors.
* Replace MIT SPDX full license name with short identifier.
* Fix sub bin calculation in mkldnn plugin
Co-authored-by: jdanieck <jozef.daniecki@intel.com>
* Refactor backend tests
* Rewrite reference implementation
* Align ref impl signature with order of inputs in v5
* Remove legacy code for training and backprop
* BarchNorminference operation class review
* Use reference to const in helpers to validate input shapes
* Refactor type_prop tests to cover v0 and v5
* Add type_prop tests
* epsilon attribute test with invalid value
* invalid integer input types
* Add serialization single layer test
* Add attribute visitor test
* Fix arm plugin test failure with dynamic element type
* Remove CoordinateTransform index call
* Add attribute count to visitor test
* Enabled Wall for nGraph
* Fixed Linux build
* Fixed ONNX Editor and class-memaccess
* Fixed topK
* Fixed some builds
* Fixed nGraph Python API
* Use std::fill instead of cycle
* Wno-unded and deprecation for app python versions
* Add deformable_convolution unit tests
* add 2D unit test
* Add unit-tests for 3D deformable_convolution
* create shared test class for deformable_convolution
* create single layer test for cpu plugin
* create single layer test for serialization
* add deformable_convolution to opset1
* create empty evaluate method
* add group and deformable_group to parameters
* Create impl for generateInput method
* add deformable_val to deformable_conv paramset
* create additional unit-tests
* Remove 3D single layer tests
* Update GenerateInput and SetUp method
* Update parameters values for SLT
* Update GenerateInput method and create Validate method
* Create additional parameters set for cpu plugin SLT
* Create unit-test with padding and add it to disabled
* add interpreter unit tests to manifest
* style-apply
* create file for reference impl
* Update year of copyright and make Validate an override method
* Update parameters names for serialization single layer tests
* Update parameters names for functional single layer tests
* add failing unit tests for interpreter to manifest
* make tests parameters more readable
* style-apply
* Include deformable_convolution inside evaluates_map
* add support for groups parameter for reference impl
* style-apply
* remove DeformableConvolutionParams struct
* fix bug with filter
* fix bug with offset type
* Update interpreter manifest
* impl evalute method
* style-apply
* Update year
* Update test names
* add utils func for deformable convolution operation
* fix filter group count
* add calculation of index of offset
* Update offsets in unit test
* add support for multiple input channels
* add padding tests
* add padding support
* style apply
* update copyright year
* create validation check helper
* Update convolve function name
* update copyright year
* style-apply
* remove integer type from serialize layer tests
* add tests for deformable_groups
* fix bug with group > 1
* style-apply
* add group unit test
* create additional group tests
* fix bug with groups attribute
* Enhance dynamic shape inference of validate and infer types method
* Add type_prop unit tests
* Fix broken op create test in python api
* fix bug with shapes in group tests
* update deformable_convolution method
* add unit test with 2 groups and 2 defromable groups
* Fix code style.
* Update UT manifests with current test status.
* Refactored backend test: names, removed duplication, add TOODs
* Add missing test cases in 'integral offsets' group.
* Fixed group attribute.
* Update interprer manifest to disable tests with integer offsets.
* Fix style.
* Remove changes in operator class.
* Revert "Enhance dynamic shape inference of validate and infer types method"
This reverts commit 2f9ce2ccd4.
* Revert "Add type_prop unit tests"
This reverts commit 944af98b8c.
* Revert "Fix broken op create test in python api"
This reverts commit 72fbfc2967.
* Fix op class.
* Convert implementation to 2D.
* Simplify implementation.
* Fix centos build.
* Reimplemented offsets handling.
* Fixed integral offsets test cases.
* Fixed deformable group attribute.
* Add bilinear interpolation.
* Refactoring regarding tolerance_bits.
* Fix groups & def_groups test case.
* Add more unit tests for group & defgroup attribute.
* Remove debug code.
* Minor refactoring.
* Add integer types to SLT.
* Revert "Add integer types to SLT."
This reverts commit 2fefe8926d.
* Add tests with real number offsets.
* Refactored bilinear_interpolation().
* Turned on SLT.
Additionally refactored and offset input set to range <0,2>.
* Update headers with short version.
* Fix SLT offests generation with int offsets.
* Add integer types to SLT.
* Fix grup + def_group test case.
* Add ticket to address IE_CPU backend test failures
* Enable real resolution for deformable values
* Add op to list of trusted operations in python script
* Fix comparison of integer expressions of different signedness compilation error
* Add comment with closing namespace and empty lines
Co-authored-by: pszmel <piotr.szmelczynski@intel.com>
Co-authored-by: ggalieroc <gabriele.galiero.casay@intel.com>
* clean SquaredDifference op from legacy FusedOp operation
* add backend and type_prop tests for squared_difference operation
* apply requested changes
* fix op name
* revert date change
* update squared_difference op info with ngraph RTTI
* change test since squred_diff should be binop
* Started to write the header file for (I)DFT reference implementation.
* Continued to write the header file for (I)DFT reference implementation.
* Renamed the header file for the reference implementation of (I)DFT.
* Started to write an implementation file for the reference implementation of (I)DFT.
* Continued to write an implementation file for the reference implementation of (I)DFT.
* Continued to write an implementation file for the reference implementation of (I)DFT.
* Small fix.
* Written copying data from input and copying data to output.
* Code style fixes.
* Small fix.
* Some fixes.
* Some fixes.
* Small fix.
* Written naive version of (I)DFT calculation.
* Some fixes.
* Some fixes.
* Some fixes.
* Some fixes.
* Some fixes.
* Small fixes.
* Written the draft of the reference implementation of (I)DFT.
* Small fix.
* Small fix.
* Code style fixes.
* Added evaluation of (I)DFT to evaluation_map.cpp.
* Small fixes.
* Some fixes.
* Written test for evaluation of 1D DFT.
* Fixed ngraph/test/CMakeLists.txt.
* Disabled DFT evaluation test in CPU, because (I)DFT has not implemented yet in CPU.
* Added debug prints to evaluation test of DFT.
* Added debug prints into evaluate_map.cpp for DFT evaluation.
* Added some debug prints into FFT calculation.
* Added more debug prints.
* Added more debug prints.
* Added more debug prints.
* Added more debug prints.
* Added more debug prints.
* Small fix.
* Added more debug prints.
* Added more debug prints.
* Small change.
* Some fixes.
* Small fix.
* Some changes.
* Added more tests.
* Added test for IDFT 1D calculation.
* Some fixes.
* Added more debug prints.
* Small fix.
* Small fix.
* Some fixes.
* Some fix.
* Small fix.
* Added tests for 2D case of IDFT.
* Some fixes.
* Written tests for 3D case of DFT.
* Some fixes.
* Small fix.
* Added test for 3D case of IDFT.
* Some fixes.
* Deleted debug prints from tests for IDFT.
* Deleted debug prints from tests for DFT.
* Deleted debug prints from the reference implementation of (I)DFT.
* Code style fixes.
* Deleted debug prints from evaluates_map.cpp.
* Written the header file for the base class of DFT and IDFT operations.
* Written an implementation of the base class of DFT and IDFT.
* Now nGraph IDFT operation class is a derived class of FFTBase.
* Now the nGraph operation DFT is a derived class of FFTBase.
* Added assert for axes in (I)DFT reference.
* Small refactoring.
* Deleted commented code.
* Small refactoring.
* Small fix.
* Initialization of calculations of the reference implementation of (I)DFT was moved in the separate function.
* Small fix.
* Code style fix.
* Small fix.
* Now evaluate() of (I)DFT uses canonicalize_axes() from the reference implementation.
* Code style fix.
* Deleted commented code.
* Added tests for i32 axes of DFT.
* Added test for i32 axes of 2D DFT.
* Added i32 axes case to test for 3D DFT.
* Added test for i32 axes in tests for IDFT.
* Written signal_size case test for 1D DFT.
* Small fix.
* Written test for signal_size case for 2D DFT.
* Added test for bfloat16 input data of 1D DFT.
* Small fix.
* Small fix.
* Small fix.
* Some fixes.
* Some fix.
* Added bfloat16 input tests for 2D DFT.
* Some fixes.
* Written tests for bfloat16 input of 3D DFT.
* Some fixes.
* Some fixes.
* Added tests for bfloat16 input of 1D IDFT.
* Some fixes.
* Added tests for bfloat16 input of 2D IDFT.
* Added test for bfloat16 input of 3D IDFT.
* Small fix.
* Some fixes.
* Added tests for float16 input of 1D DFT.
* Small fix.
* Written tests for float16 input of 2D and 3D DFT.
* Small fix.
* Some fixes.
* Some fixes.
* Written tests for float16 inputs of 1D, 2D, 3D IDFT.
* Some fixes.
* Some fixes.
* Some fixes.
* Some fixes.
* Deleted redundant include.
* Some fixes.
* Added tests of 1D and 2D DFT for the case when some axes lengths are powers of 2.
* Added tests for 3D DFT and 1D, 2D, 3D IDFT in the case when lengths of some axes are powers of 2.
* Small fix.
* Added some comments.
* Added some comments.
* clean the FusedOp from mod operation
* add backend and type_prop tests for mod operator
* convert taking autobrodcast to match binary elementwise arithmetic ops
* add type_prop/mod.cpp to CMakeLists.txt
* fix style
* fix style v2
* remove evaluate method and add backend test for negative numbers
* add copyright for type_prop/mod.cpp
* Added reference implementation for Roll operation.
* Small corrections.
* Removed duplicate test disabling.
* Changed implementation using manual data manipulation.
* Removed unnecessary function.
* Corrected tests, added converting axes and shift to int64.
* Auto benchmarks for evaluate(...) and extra unit tests.
* Test suite for manual i/o examination.
* Template on input type. Constant folding tests the 'count_ops_of_type' issue.
* Multidim manual tests
* NgOp&Ref: Unit tests Ok. NB evaluates_map
* Stable. Negative indices as if larger than depth. Benchmarks removed.
* Added Serialization SLT
* I/O shapes consistency checks
* Consistency checks in evaluate(). Removed the ref from evaluates_map.
* removed excessive #include from evaluates_map
* Code style
* Add BinaryConvolution unit tests.
* Changed types to u1.
* Add BIN precision handling in TestCase class.
* Refactored validate and infer types to enhance dynamic shape inference
* Add type_prop test to cover invalid op cases and dynamic shapes
* Fix style
* Disable check for float type of data batch input
* Add type_prop test for incompatible input channels in inputs
* Disable backend unit tests
* Fix style
* Add reference implementation
* Add backend tests
* Add single layer tests
* Add check for float element type of batch data input
* Refactor backend test cases to compare with regular convolution
* Add serialization tests
* Clean up
* Add 1D and 3D tests into op_eval
* Changes in reference implementation to improve readability
* Add ticket information for todo tasks
* Fix implementation misbehavior for filter channels
* Add backend unit tests to cover strides, dilations, padding, channels and batches
* Add end of line into files
* Change name of type_prop unit tests
* Simplified lambda to get spatial dimensions of filters
* Add comment to support filters input as Parameter
* Add namespace details for BinaryConvolution utility functions
* Address review comments
Co-authored-by: jdanieck <jozef.daniecki@intel.com>
* Added support for Gelu-6 to the MO
* Adding Gelu-6 to ngraph and python API + some tests
* Fixed typo in the Gelu approximation mode
* Fixed Gelu-6 reference implementation for Tanh mode
* Added transformation to downgrade v6::Gelu to v2::Gelu
* Added specification for the Gelu-6
* Code style fixes
* The Gelu-6 operation specification update
* Fixed compilation issue in reference implementation for Gelu
* Fix compilation issues for some OSs
* Code style fix
* One more cpplint issue fix
* Fixed Gelu6 reference implementation compilation on Windows.
* Code style fix
* Fixed various ngraph unit tests
* Code style check
* Reverted Gelu-2 to be fused op
* Fixed Gelu6 downgrade transformation
* Added unit test for Gelu6Downgrade transformation
* Update copyright year
* Updated copyright year
* Replaced tab characters with 4 spaces in IR reader tests
* Code style fixes
* Added default value for GeluApproximation mode for Gelu-6 op
* Fixed code style for Gelu-6
* Changed order of parameters for the Gelu evaluate to potentially avoid backward compatibility issues with ARM plugin
* Fixed code style
* Introduced opset7. Moved Gelu6 to opset7
* Fixed non-updated transformation
* Fixed opset version in ngraph Python API for Gelu operation
* Fixed typo in the opset number in the documentation
* Reverted some changes related to Gelu6
* Updated MO to produce Gelu7
* Updated unit tests for Gelu
* Updated Gelu7 specification
* Changed gelu reference implementation. Added opset7 to Python packages
* Updated Python API tests for Gelu operation
* Code style fix
* Marked get_approximation_mode function as const
* Added missing "const" qualifier
* Fixed code style issues in tests
* Added extractor for MxNet operation Gelu
* Spelling issues fix
* Updated MxNet supported symbols
* Added NGRAPH_OP_SCOPE for Gelu7 validate_and_infer_types
* Fixed a typo in the comment
* Reference implementation for Proposal, enable CPU SLT
* code style fix
* add type prop test for invalid anchor count
* add unit test
* fix shapes in attribute test
* temp workaround- disable maring end of boxes list
* Disable CPU smoke test- spec misalignment
* code style fixes
* add some details to the specification
* disable myriadx proposal slt
* review changes, using usigned int and size_t
* improve proposal op shape inference to cover dynamic too, add unit test coverage
* remove unused variable in test body
* remove batch size in tests where its not used
* add post nms topn initialization in tests where it was missing
* review comments
* style fix
* style fix 2
* add tests, remove unused variables, change shape inference checks
* style fix
* add input tensors type checks and test coverage
* align input type in attribute and ngraphreader tests to match specification
* fix wrong dimension in error message
* proposalv4 ref impl
* enable single layer and unit tests for proposalv4 ref impl
* align output termination with cpu, enable cpu slt
* custom slt compares to detect less-than-predicted number of boxes
* custom slt compares to detect less-than-predicted number of boxes
* Clarify output termination in spec
* review comments
* smaller input data for unit tests
* add check for batch_dim being static
* disable gpu slt for proposal
* test data style fix
* test data style fix 2
* add type section to specification
* shape inference improvement
* multiply expected 1st dim in tests by post_nms_topn
* add checks and testcases for dynamic ranks
* indentation, review comments
* reduce code redundancy in ref implementation
* remove comment
* Fix typo in proposal1 spec
* Fix typo in proposal4 spec
* Move Convolution and ConvolutionBackpropData ref impls into separate files.
* Add convolution unit tests.
* New convolution reference implementation.
* Remove unused convolution ref impl argument.
* Fix style.
* Revert "Remove unused convolution ref impl argument."
This reverts commit 739065d0d0.
* WA for arm-plugin: additional include with ConvolutionBackpropData.
* Style format in Convolution SLT CPU instantiation.
* Add 1D Convolution SLT CPU tests.
* Add Convolution Serialization SLT.
* Update source banners with 2021 date.
* Specification review.
* Readability improvement in padding detection.
* Refactoring regarding Tensor usage.
* Iteration over tensor slices made more readable.
* Code refactored to use only one convolution implementation.
3D convolution is used to compute also in 1D & 2D case (parameters,
inputs and filters shapes are adjusted accordingly).
* Removed Tensor abstraction.
* Name unnamed namespace as convolution_details.
* Refactoring: replaced std::next + negative index with std::prev.
* Specification refactoring.
* Revert "Name unnamed namespace as convolution_details."
This reverts commit cea526ec49.
* Added new convolution() overload.
* Fix legacy convolution() overload (needed for kmb-plugin).
* Reduced number of template type arguments in convolution ref impl.
* Added 'output' section in Convolution spec.
* Remove floating round type configuration.
* ConvertLike: Develop reference implementation
* ConvertLike: Enable single layer tests for GPU plugin
* ConvertLike: Enable bf16 precision for evaluate method
* ConvertLike: Add unit tests
* ConvertLike: Add dynamic shape test case
* ConvertLike: Remove unnecessary ngraph namespace and using declaration for v1::ConvertLike
* ConvertLike: Simplified reference::convert by using std::enable_if
* Bucketize: Revise op class and add type_prop unit tests
* Bucketize: Develop reference implementation
* Bucketize: Add unit tests
* Bucketize: Add single layer test and cpu instantiation
* Bucketize: Add unit test with empty buckets for INTERPRETER
* Bucketize: Typo in buckets element type check
* Bucketize: Add custom generated inputs in single layer test class
* Bucketize: Use random_device to generate seed for data blob
* Bucketize: Remove unsupported f64 precision
* Bucketize: Add function description
* Bucketize: Remove randomness of inputs generation by using static seed
* Bucketize: Support different precisions for data and bucket inputs
* Bucketize: Refactor type_prop tests and improve backend unit test coverage
* Revice DetectionOutput reference implementation
Ticket: 37433
* fix test_create_op
* fix test_dyn_attributes
* apply code format
* fix crash on Windows when variance_encoded_in_target == 1
* add more checks for DetectionOutput inputs
* Fix single layer tests
* apply code format
* fix ssd_vgg16_300 inference with batch size > 1
* update types in docs
* fix crash on windows
* apply code style
* fix python tests
* fix setting output type
* change False to false and True to true in docs
* Allow prior boxes to have different type than box logits
Some models work that way
* simplify output shape calculation
* fixes to docs
* Add reference implementation for PSROIPooling operator
* fix test_roi_pooling
* use std::roundf
* remove unnecessary copies in single layer tets
* Fixes after review
* fixes after review
* use element::Type_t instead of element::
* apply code format
* add PSROIPooling to evaluates_map
* apply code format
Gabriele Galiero Casay