+ Prevent to fuse PRelu if convolution is onednn impl.
+ Prevent to remove input reorder if onednn conv converts format from bfyx to b_fs_yx_fsv32.
+ Add a value to optimal lws values for eltwise kernel selector.
+ Fix a condition of remove redundant reorder. (try_fuse_reorder_bfyx_to_fsv32)
* [GNA] More precized calculation of Log pwl and pwl with sf < 1
* [GNA] Added tests
* [GNA] Types correction
* [GNA] Added comment for absolute threshold calculation in the test
* Deprecated old evaluate methods
* Suppress node
* Fixed some warnings
* Fixed documentation and warnings
* Fixed some warnings
* Disable some warnings
* work with omz: change hardcoded paths for memory_tests, fix paths for memcheck get_testdata.py script
* change paths to omz tools
* revert mo changes
* remove mo from arguments
* remove extra var from cmd string
* remove is_db_used check
* add timeout to proc_exec
* merge master
* Fix performance on resnet50 quantized models
LP transformations won't work on the model unless the last 4 inputs to FakeQuantize
are constants. In order to meet that requirement, we need to perform constant folding
for those inputs in QuantizeLinear ONNX operator.
Ticket: 65375
* fix "Cannot find blob with name: y" exception during onnx_model_quant_conv_linear
* remove linking with onnx_ngraph_frontend
* fix exclude path
* Remove fp16 of Convert layer test from skip_tests.config.cpp as it works now
* update repo
* add convolution reference test and ngraph backend test
* fix location of ngraph visitor convolution.cpp
* Pre-process:
- Implicit conversions for element type and layout
- 'convert_element_type' with default argument to network
- Convert_element_type - don't add ops if dst and src types are same
- Convert_layout - don't add ops if dst and src layouts are same
- Custom step - use Output<Node> instead of shared_ptr<Node>
- Support of addressing input by tensor name
Post-process:
- Avoid duplication of tensor names after post-processing
* Fixed IE tests
* Enabled support of IR v11 in IR frontend
* Enabled serialization to v11
* Added indexes for Inputs/Outputs
* Fixed typo
* Fixed code style
* Fixed deserialization tests
* FIxed typo
* Avoid changes in the original serialized function
* Split IRv7 and IRv10 code
* Added preprocessing steps to read network
* Added preprocessing for new API for IR v10
* clang-format
* Updates for compiler flags
* Reverted some changes
* Fixed TI tests
* Removed version from the function and indexes from parameters/results
* Restore function version
* Clone function rt info
* Removed v11 version
* Added serialization of old API map in ngraph.
* Changed order type to int64_t.
* Fixed uint64_t error, added comments.
* Apply suggestions from code review
Co-authored-by: Gleb Kazantaev <gleb.nnstu@gmail.com>
* Read old_api_map
* Added tests with undefined type and empty order.
* Fixed reprocessing
* Updated gitignore
* Updated serialize / deserialize tests
* Added set, get and has methods.
* Fix in tests.
* Applied old_api_map tests
* Fixed empty convert_layout
* Allow Layout from empty string
* Fixed clang-format
* Export FEM::load_by_model
* Apply suggestions from code review
Co-authored-by: Ilya Churaev <ilyachur@gmail.com>
* Made inline methods, changed to shared_ptr.
* Small fix.
* Moved methods to header file.
* Small fix.
* Applied post-processing
* Fixed review comments
Co-authored-by: Ilya Churaev <ilya.churaev@intel.com>
Co-authored-by: Anastasia Popova <anastasia.popova@intel.com>
Co-authored-by: Gleb Kazantaev <gleb.nnstu@gmail.com>
Co-authored-by: Ilya Churaev <ilyachur@gmail.com>
* Added ConcatAlignFilter support for Accuracy Aware mode
* fails after reviewing and added tests
* moved the test to gna/scale_factors_tests; added comments to test
* Added serialization of old API map in ngraph.
* Changed order type to int64_t.
* Fixed uint64_t error, added comments.
* Apply suggestions from code review
Co-authored-by: Gleb Kazantaev <gleb.nnstu@gmail.com>
* Added tests with undefined type and empty order.
* Added set, get and has methods.
* Fix in tests.
* Apply suggestions from code review
Co-authored-by: Ilya Churaev <ilyachur@gmail.com>
* Made inline methods, changed to shared_ptr.
* Small fix.
* Moved methods to header file.
* Small fix.
Co-authored-by: Gleb Kazantaev <gleb.nnstu@gmail.com>
Co-authored-by: Ilya Churaev <ilyachur@gmail.com>
