Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 16 Jan 2019 16:45:39 +0000 (19:45 +0300)]
[nnc] Add temporary im2col buffer reuse (#2859)
Add temporary buffer reuse to soft backend for Conv
Implemented via a temporary tensor
Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 16 Jan 2019 16:12:19 +0000 (19:12 +0300)]
[nnc] New importers/style transfer ops support (#2705)
Added resize to caffe2 and ONNX (as upsample)
Tested with 1 layer nets.
Signed-off-by: Andrei Shedko a.shedko@partner.samsung.com
[nnc] The initial implementation of Shape op in ONNX (#2868)
inception_v3 model requires this operation.
In addition I changed the includes order inside some ONNX files.
Signed-off-by: Andrew V. Tischenko a.tischenko@partner.samsung.com
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 16 Jan 2019 14:22:21 +0000 (17:22 +0300)]
[nnc] Remove BiasAdd and ScaleOp from Add operation in Caffe2 frontend (#2857)
* Remove BiasAdd and ScaleOp from caffe2 convertAdd
* Fix BatchNorm assertion of caffe frontend
* Fix Elementwise with broadcast on caffe frontend
Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
[nnc] Mobilenet support in soft backend in ONNX (#2866)
Soft backend can't load int64 constant tensors that's why it was transformed in float one (reshape operation).
Signed-off-by: Andrew V. Tischenko <a.tischenko@partner.samsung.com>
[nnc] Make FullyConnectedOp treat the second tensor as ordinary argument (#2858)
* Change the signature of FullyConnectedOp to identically handle both input parameters.
* Refactor uses of FullyConnectedOp.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Some ops fixes, mobilenet support on ONNX. (#2856)
There were fixed/implemented the following operators:
PadOp, GivenTensorFillOp, ConstantOp, ConvOp.
As result we have now 2 working NNs: resnet50 and mobilenet. The last one was tested on interpreter only.
Signed-off-by: Andrew V. Tischenko <a.tischenko@partner.samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 16 Jan 2019 06:53:55 +0000 (15:53 +0900)]
[enco] Fix comment for duplicate_inout_bag (#2864)
This will fix a typo and explanation of the function
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 15 Jan 2019 23:55:08 +0000 (08:55 +0900)]
[enco/tflite] Add Regression 0003 testcase (#2854)
This commit introduces Regression 0003 testcase whose inputs are all
constant.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
[nnc] Adapt ONNX importer to modified ScaleOp interface (#2860)
Take into account modified interface of ScaleOp in ONNX importer.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Make BiasAddOp and ScaleOp treat the second tensor as ordinary argument (#2848)
* Change the signatures of BiasAddOp and ScaleOp to identically handle both input parameters.
* Refactor uses of BiasAddOp and ScaleOp.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Refactor TFLite importer to support operators with multiple outputs (#2816)
Make operator conversion methods accept and return vectors of `IODescriptor`s.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 15 Jan 2019 10:23:04 +0000 (19:23 +0900)]
[tflchef] Support Sqrt operation (#2695)
This commit supports generating tflite file of Sqrt operation.
Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 15 Jan 2019 06:15:05 +0000 (15:15 +0900)]
[tflchef] tlchef for FullyConnected (#2852)
* [tflchef] tlchef for FullyConnected
This commit adds code to convert recipe for FullyConnected to tflite file.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* alphabetical order
* remove comment about alphabetical order
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 15 Jan 2019 03:56:25 +0000 (12:56 +0900)]
[enco] Refactor DuplicatePass (#2850)
This will hide DuplicatePass class and provide duplicate_inout_bag() like the other transformation pass
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 15 Jan 2019 03:54:29 +0000 (12:54 +0900)]
[enco] Refactor SplitPass (#2849)
This will hide SplitPass class and provide split_into_phases() like the other transformation pass
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 15 Jan 2019 02:07:39 +0000 (11:07 +0900)]
[enco] Comment for bypass_shuffle (#2843)
* [enco] Comment for bypass_shuffle
This will add comment for generate_bypass_shuffle method
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* apply comments
* apply comments
* apply comment
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 15 Jan 2019 01:57:30 +0000 (10:57 +0900)]
[coco] Add comment of Bag Input Output (#2846)
* [coco] Add comment of Bag Input Output
This will add explanation comment for input and output attributes
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* apply comment
[nnc] Refactoring cmake build system (#2844)
* make separate config and utils files
* refactoring CMakeLists in other components
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
[nnc] Refactor ONNX importer to support operators with multiple outputs (#2831)
Make operator conversion methods accept and return vectors of IODescriptors.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Fix soft backend implementation for Gemm (#2750)
- Fixed generation of cpu soft backend gemm operation
- Small refactoring and typo fixes of interpreter and onnx frontend
Signed-off-by: Andrew V. Tischenko <a.tischenko@partner.samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 14 Jan 2019 07:48:55 +0000 (16:48 +0900)]
Integrate TensorFlow Lite 1.12 (#2838)
* Integrate TensorFlow Lite 1.12
This commit introduces the following changes to use TensorFlow Lite 1.12
instead of TensorFlow Lite 1.7
- Use the latest Flatbuffers (v1.10)
- Update Flatbuffers Build methodology
- Update nnkit's tensorflow lite support library
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Fix a typo (avaialable -> available)
* Remove debug code
* Update status message
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 14 Jan 2019 02:18:18 +0000 (11:18 +0900)]
[enco] Support constant folding for add, sub, mul, div, sqrt (#2821)
* [enco] Support constant folding for add, sub, mul, div, sqrt
Until now, only `coco::Copy` instruction was available for constant folding.
This commit will support some `coco::Eval` instructions in constant folding,
which is add, sub, mul, div, sqrt.
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* minor modifications
[nnc] Add check for include generation correctness in acl unit tests (#2722)
- Add checks for correct include DOM generation
- Fix system headers setter in ArtifactModule
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Fri, 11 Jan 2019 20:54:33 +0000 (23:54 +0300)]
[nnc] Made importer error message prettier (#2761)
Made importer error message more prettier
Message example:
```
NNC can't load model. Detected problems:
* Cos: unknown layer
* Only 'NCHW' axis order is supported
* Sin: unknown layer
```
Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
[nnc] Remove transposes of inputs and outputs in importers (#2837)
Remove transposes of input and outputs tensors in Caffe and Caffe2 importers
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Remove the debugTranspose option (#2813)
Remove the `debugTranspose` option which is no longer needed.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Generate getters for unnamed output operations (#2830)
- Remove redundant type in operation analysis
- Move output nodes registration outside of operation analysis
- Refactor artifact tensor declaration in soft backend
- Create persistent tensor for output node without name
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
[nnc] Bugfix for the SVACE warning (#2743)
Pass large arguments by reference and small by value
Signed-off-by: Timur Ablyazimov <t.ablyazimov@samsung.com>
[nnc] Free temporary tensors to reduce memory consumption of ACL artifact (#2767)
- Remove redundant reverse shape transposes like this [0,1,2,3]->[3,2,1,0]
- Refactor tensor serialization
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 11 Jan 2019 08:28:54 +0000 (17:28 +0900)]
[enco] Comment for dead object elimination (#2819)
This will add a comment for getting dead objects
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 11 Jan 2019 08:05:41 +0000 (17:05 +0900)]
Introduce TensorFlow Lite 1.12 package (#2820)
This commit introduces TensorFlow Lite 1.12 package.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 11 Jan 2019 06:37:09 +0000 (15:37 +0900)]
Introduce Abseil package (#2818)
This commit adds Abseil package which provides abseil target.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 11 Jan 2019 00:38:45 +0000 (09:38 +0900)]
[encodump] Support Div operation (#2814)
This commit would enable encodump to get model with Div operation
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 11 Jan 2019 00:37:51 +0000 (09:37 +0900)]
[nnsuite] Use specific tflite version (#2807)
TensorFlow Lite currently has no public API, and thus its
interface(=API) is subject to change.
Let's use specific tflite version to avoid compilation failure due to
API incompatibility.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 11 Jan 2019 00:19:49 +0000 (09:19 +0900)]
[enco] Add comments for constant folding (#2808)
* [enco] Add comments for constant folding
This will add some comments of constant folding
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* apply suggestion
* use capital letter
* fix grammer
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 11 Jan 2019 00:19:29 +0000 (09:19 +0900)]
[enco] Add test case for Div operation (#2812)
This commit adds test case for Div support in enco
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 10 Jan 2019 15:40:22 +0000 (18:40 +0300)]
[nnc] Fix incorrect operation id for leaky relu (#2815)
Id of LeakyRelu operation was incorrectly set as ReLU
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
[nnc] Check that the model has the correct format in Caffe importer (#2753)
Ensure that the model file consumed entirely, otherwise complain that the model cannot be loaded. Fixes #2050.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 10 Jan 2019 13:04:34 +0000 (16:04 +0300)]
[nnc] Support Clip and Reshape operations on caffe2 importer (#2751)
* Support GivenTensorInt64Fill const tensors
* Convert Clip operation in CappedRelu
* Support Reshape operation
Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
[nnc] remove redundant shape transposes from acl backend (#2764)
- Remove redundant reverse shape transposes like this [0,1,2,3]->[3,2,1,0]
- Refactor tensor serialization
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 10 Jan 2019 09:30:28 +0000 (12:30 +0300)]
[nnc] Fix deconv2D import in tflite importer (#2544)
Use output shape stored in tflite model as it's impossible to compute it from operation parameters
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 10 Jan 2019 09:26:33 +0000 (18:26 +0900)]
[enco] Support Div for tflite frontend (#2779)
* [enco] Support Div for tflite frontend
This commit extends enco tflite frontend including Div operation
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Rename ifm, ofm indices to 'num', 'denom' and 'out'
* Add activation assertion
* Input shapes may differ in case of broadcasting
* Remove redundant variable d
* Add TODO notice about object type generalization
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 10 Jan 2019 09:16:10 +0000 (18:16 +0900)]
[ann] Fix Div operation to perform division (#2806)
This commit fixes Div operation in ann to correct way
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 10 Jan 2019 08:16:53 +0000 (17:16 +0900)]
Support Pre/Post Package Hook (#2798)
With this commit, nncc_find_package invokes pre (and post) hooks before (and after)
actual find_package invocation.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 10 Jan 2019 04:46:55 +0000 (13:46 +0900)]
Introduce nnkit_support_tflite-1.7 target (#2804)
This target allows clients to specify the version of tensorflow lite that
they want to use.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 10 Jan 2019 04:16:14 +0000 (13:16 +0900)]
[ann] Support Div operation (#2800)
* [ann] Support Div operation
This commit would support Div operation in ann
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Change indent and minor refactoring
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 10 Jan 2019 04:08:54 +0000 (13:08 +0900)]
Introduce Abseil Source package (#2802)
* Introduce Abseil Source package
This commit introduces Abseil (C++) source package as a nncc external
package.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Update comment
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 10 Jan 2019 04:07:23 +0000 (13:07 +0900)]
[enco] fix typo in comment (#2801)
This will fix typo in reduce_identical_object comment
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 10 Jan 2019 02:17:51 +0000 (11:17 +0900)]
Update external dependencies on URL change (#2799)
With this commit, ExternalSource_Download will re-download external source
if its URL has changed.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 10 Jan 2019 02:10:52 +0000 (11:10 +0900)]
Add .vs folder to gitignore (#2797)
This will add to ignore Visual Studio temporary .vs folder
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 10 Jan 2019 01:05:01 +0000 (10:05 +0900)]
Remove deprecated TensorFlowLite package (#2786)
Let's use versioned package instead.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 10 Jan 2019 01:04:46 +0000 (10:04 +0900)]
[coco] Add comment of Locatable (#2796)
* [coco] Add comment of Locatable
This will add comment for Locatable interface
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* simplify
[nnc] Support transpose in acl backend (#2746)
- Support transpose operations in acl backend
- Insert transposes in acl backend where needed
- Add test for transpose operation
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
[nnc] Support for LSTM Caffe layer (#2721)
* Add support for LSTM Caffe layer.
* Rename "SoftmaxLoss" Caffe layer to "SoftmaxWithLoss" as it should be.
* Add support for non-default axis for InnerProduct Caffe layer.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 9 Jan 2019 11:46:20 +0000 (14:46 +0300)]
[nnc] Add LeakyRelu activation (#2757)
* [nnc] Add LeakyRelu activation
Implement leaky relu activation function( relu with negative slope )
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 9 Jan 2019 10:57:44 +0000 (19:57 +0900)]
[enco] Fixed debugging typo (#2793)
enco_dump_all_bags, _objects, _ops, _instrs take module as a parameter.
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 9 Jan 2019 08:42:12 +0000 (17:42 +0900)]
[enco] Support Div for ANN backend (#2792)
* [enco] Support Div for ANN backend
This commit enables ANN backend generation for Div operation
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Remove irrelavant modification in 'Debugging.cpp'
Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 9 Jan 2019 08:29:33 +0000 (11:29 +0300)]
[nnc] Fix for constant in ModelAnalyzer (#2744)
Fixed constants not being visible in soft backend
Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 9 Jan 2019 08:14:16 +0000 (17:14 +0900)]
[enco] add comments for loads in DataLayoutConversion (#2789)
* [enco] add comments for loads in DataLayoutConversion
This will add some comments explaining the condition
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* remove 'deleted'
* add about bounded
* use 'note'
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 9 Jan 2019 04:25:53 +0000 (13:25 +0900)]
[coco] Support Sqrt operator in coco (#2790)
This commit adds Sqrt op in coco.
Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 9 Jan 2019 00:21:40 +0000 (09:21 +0900)]
Introduce TensorFlow 1.12 source package (#2785)
This commit integrates TensorFlowSource (1.12) package into nncc build
system.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 8 Jan 2019 23:41:56 +0000 (08:41 +0900)]
[enco] Add comments to subst method (#2775)
* [enco] Add comments to subst method
This will add explanation comment of subst() method
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* some changes and example code
* use word USER
* Apply suggested comment
* fix typos
* use word Consumer
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 8 Jan 2019 08:44:44 +0000 (17:44 +0900)]
[nnkit] Use specific version of TensorFlow Lite (#2782)
This commit updates nnkit_support_tflite uses a specific version of
tensorflow lite (v1.7).
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 8 Jan 2019 08:38:46 +0000 (17:38 +0900)]
ExternalSource_Download with URL keyword (#2784)
This commit updates ExternalSource_Download to accept URL keyword as an
argument of the following form:
ExternalSource_Download(SAMPLE DIRNAME "SAMPLE" URL "...")
This change will make it easy to understand CMake build script.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 8 Jan 2019 07:24:20 +0000 (16:24 +0900)]
Add TensorFlow Lite 1.7 package (#2778)
This commit introduces TensorFlow Lite 1.7 package to nncc build system.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 8 Jan 2019 00:49:54 +0000 (09:49 +0900)]
[coco] Support Div operation (#2770)
This commit extends coco IR including Div operation
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 3 Jan 2019 01:29:41 +0000 (10:29 +0900)]
[enco] Add Sub test case (#2713)
This commit add Sub test case for enco
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 2 Jan 2019 05:47:47 +0000 (14:47 +0900)]
[encodump] Add Sub operator (#2731)
This commit supports to print information about Sub.
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 2 Jan 2019 04:15:42 +0000 (13:15 +0900)]
[tflchef] Support Div operation for reverse (#2760)
This commit will enable Div operation in tflchef-reverse
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 2 Jan 2019 00:08:49 +0000 (09:08 +0900)]
[enco] Add cassert headerfile (#2749)
Add cassert header file in operator cpp files
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 2 Jan 2019 00:08:36 +0000 (09:08 +0900)]
[RefNN] Remove commented code (#2748)
Remove unnecessary comment for readability.
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 2 Jan 2019 00:08:17 +0000 (09:08 +0900)]
[tflchef] Support Div operation (#2755)
This commit will let tflchef cook Div operation
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
[nnc] Add more tests for acl dom to text operations (#2719)
Tested entities:
- ArtifactIndex
- ArtifactRet
- ArtifactBreak
- ArtifactCont
- ArtifactVariable
- ArtifactBlock
- ArtifactForLoop
- ArtifactIf
- ArtifactFunction
- ArtifactClassVariable
- ArtifactClassFunction
- ArtifactClass
- ArtifactModule
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 26 Dec 2018 14:18:41 +0000 (17:18 +0300)]
[nnc] Fix interpreter deconv implementation (#2756)
Rotate kernel index prior to input index calculation
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
[nnc] Support constant in acl backend (#2723)
- Support generation of constant operation in acl backend
- Add test for constant operation
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 24 Dec 2018 08:21:06 +0000 (17:21 +0900)]
TensorFlowSource v1.7 package (#2715)
This commit adds "version" property to TensorFlowSource CMake package
in nncc.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 24 Dec 2018 06:03:12 +0000 (15:03 +0900)]
[ann] Implement Sub (#2714)
* [ann] Implement prepare for Sub
This patch implements a prepare for Sub operation in ann backend
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
* [ann] Implement float32 operation for Sub
This commit implements a float32 operation for Sub operator
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Fri, 21 Dec 2018 13:07:31 +0000 (16:07 +0300)]
[nnc] Caffe2 importer unittest (#2735)
Implemented caffe2 importer unittest for unsupported op
Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
[nnc] Merge frontend libraires (#2737)
* add one frontend library as interface
* remove redundant `target_include_directories`
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
[nnc] The first ONNX model resnet50 works on NNC interpreter (#2718)
Several operators were fixed: BatchNormalization, Reshape, Gemm and Pooling. Now NNC is available to convert the ONNX resnt50 network, play it back in interpreter and to produce the out which is totally comparable with reference data.
Signed-off-by: Andrew V. Tischenko a.tischenko@partner.samsung.com
Make onnxproto as static library (#2738)
* make onnxproto as static library
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
[nnc] Support sigmoid operation in acl backend (#2739)
- Generate code for sigmoid activation function
- Add unit tests for sigmoid, tanh and capped_relu activation functions
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 20 Dec 2018 17:51:03 +0000 (20:51 +0300)]
[nnc] Sub and Squared Diff (#2679)
- Refactored tflite import of various elementwise ops
- Added support for sub and SquaredDiff to C++ SB and Interpreter
- Fixed assert in c++ CPU artifact
Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
[nnc] Do not transpose weights when serializing FullyConnected operation for x86-spp back end (#2736)
Remove call to transposeTensor<1, 0> when serializing weights tensor for FullyConnected op for x86-cpp backend.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Fix buid target name (#2740)
* correct target name in CMakeLists.txt
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
[nnc] Cleanup dead code in interpreter backend (#2734)
Removed no longer used variable `_opByName` and methods working with it.
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
[nnc] Refactor ONNX importer and fix ONNX importer build bug (#2733)
* remove redundant classes from ONNX
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 19 Dec 2018 23:19:15 +0000 (08:19 +0900)]
[enco] IRValidator test code for conv2D (#2707)
* [enco] IRValidator test code for conv2D
This commit adds test code to check output shape of conv2D.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* pr fix: using std::array, removed <iostream>
* pr fix: test when validate fails
* pr fix: remove iostream
Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 19 Dec 2018 18:32:09 +0000 (21:32 +0300)]
[nnc] Added Unsqueeze to ONNX (#2729)
Added Unsqueeze to ONNX with Reshape
Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
[nnc] Create interface for NN importers (#2728)
* added factory method to create importer objects
* added template method to manage `run` method for all importers
* ONNX importer wasn't changed
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 19 Dec 2018 16:05:46 +0000 (19:05 +0300)]
[nnc] Added Resize implementation in SB (#2398)
SoftBackend implementation of Resize Nearest Neighbor
Lifted from TFLite Reference ops.
Fixed Interpreter implementation.
Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 19 Dec 2018 15:30:44 +0000 (18:30 +0300)]
[nnc] Fixed caffe2 maxpool border type (#2727)
Fixed caffe2 importer maxpool border type
Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 19 Dec 2018 12:17:20 +0000 (15:17 +0300)]
[nnc] Fix strided slice shrink axis mask (#2725)
* Fix processing shrink axis mask on StridedSlice operation creation on TFLite importer
Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 19 Dec 2018 10:50:13 +0000 (13:50 +0300)]
[nnc] Support StridedSlice operation on TFLite importer (#2720)
Support StridedSlice operation on TFLite importer only for strides array of ones
Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 18 Dec 2018 16:23:34 +0000 (19:23 +0300)]
[nnc] Fixed memory leak and heap overflow (#2703)
* Fixed memory leak and heap overflow in softbackend elementwise op
* Collect caffe2 model problems before tensors loading
Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 18 Dec 2018 13:57:06 +0000 (16:57 +0300)]
[nnc] Support Transpose operation on TFLite importer (#2716)
* Support TransposeOp on TFLite importer
* Add const where needed in TFLite op creator
Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 18 Dec 2018 07:49:13 +0000 (16:49 +0900)]
Take DIRNAME as an optional argument (#2709)
This commit refines the implementation of ExternalSourceDownload; Now,
users can pass DIRNAME as an optional argument.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 18 Dec 2018 07:43:39 +0000 (16:43 +0900)]
[enco] Add tflite frontend for Sub (#2710)
This commit will add the tflite frontend of enco for Sub.
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
윤지영/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 18 Dec 2018 07:11:41 +0000 (16:11 +0900)]
[enco] Add ann backend for Sub (#2711)
This commit will add the ann backend of enco for Sub.
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>