platform/core/ml/nnfw.git
5 years ago[moco/tf] tensor and filter for ShapeInferenceData (#3837)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 18 Jun 2019 05:11:55 +0000 (14:11 +0900)]
[moco/tf] tensor and filter for ShapeInferenceData (#3837)

This will add getter of tensor and setter of filter for ShapeInferenceData

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[tf2tflite] Use i5diff to detect shape mismatch (#3836)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 18 Jun 2019 04:08:01 +0000 (13:08 +0900)]
[tf2tflite] Use i5diff to detect shape mismatch (#3836)

It turns out that h5diff exited with zero even in the presence of
shape mismatch.

This commit updates tf2tflite test framework to use i5diff (instead of
h5diff) to detect such shape mismatch cases.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco exporter] Refactor export Transpose (#3835)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 18 Jun 2019 03:50:17 +0000 (12:50 +0900)]
[loco exporter] Refactor export Transpose (#3835)

FeatureEncode, FeatureDecode are now exported as TRANSPOSE in tflite
when they are not identity(NHWC). In future, FilterEncode will be as
well. This commit refactors exporting TRANSPOSE in a form reusable
by FilterEncode.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[locomotiv] Session setting input may throw error (#3714)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 18 Jun 2019 03:45:15 +0000 (12:45 +0900)]
[locomotiv] Session setting input may throw error (#3714)

* [locomotiv] Session setting input may throw error

Previously Session::set_input() returns false on undesired cases. With
this commit, set_input() throws runtime error on such cases.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Review feedback: amend comment

5 years agoIntroduce i5diff (#3832)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 18 Jun 2019 03:20:47 +0000 (12:20 +0900)]
Introduce i5diff (#3832)

* Introduce i5diff

This commit introduces i5diff which takes two HDF5 files (that
nnkit HDF5 exports generates) and compares them.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Enable format check

* Fix typos

5 years ago[moco/tf] remove unnecessary call to toupper (#3834)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 18 Jun 2019 03:03:18 +0000 (12:03 +0900)]
[moco/tf] remove unnecessary call to toupper (#3834)

This will remove unnecessary call to toupper of pad type string as it is already uppercase when created

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Enable Conv2D pad setter (#3833)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 18 Jun 2019 03:00:28 +0000 (12:00 +0900)]
[loco] Enable Conv2D pad setter (#3833)

This commit enable (temporarily disabled) Conv2D pad setter.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] add log for fix padding avgpool2d (#3829)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 18 Jun 2019 02:25:35 +0000 (11:25 +0900)]
[moco/tf] add log for fix padding avgpool2d (#3829)

* [moco/tf] add log for fix padding avgpool2d

This will add result pad values of fix padding of AvgPool2D node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* move LOGGER to top of function

5 years ago[moco/tf] Adding BiasAdd (#3815)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 18 Jun 2019 02:23:31 +0000 (11:23 +0900)]
[moco/tf] Adding BiasAdd (#3815)

[moco/tf] Adding BiasAdd

This comits adds BiasAdd for moco/tf.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
Signed-off-by: Pavel Iliutchenko p.iliutchenk@samsung.com
5 years ago[moco/tf] provided helper for validate methods (#3804)
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 18 Jun 2019 01:23:09 +0000 (04:23 +0300)]
[moco/tf] provided helper for validate methods (#3804)

* [moco/tf] provided helper for frontend validate method

provided helper for frontend validate method

Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
* Fixed formatting

Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
5 years ago[moco/tf] Fix padding for MaxPool2D (#3805)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 18 Jun 2019 01:17:49 +0000 (10:17 +0900)]
[moco/tf] Fix padding for MaxPool2D (#3805)

* [moco/tf] Fix padding for MaxPool2D

This will fix padding for MaxPool2D: same as AvgPool2D

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* fix formula

* fix pad formula

* fix typo

5 years ago[moco/tf] test of shape inference for avgpool2d (#3825)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 18 Jun 2019 01:09:01 +0000 (10:09 +0900)]
[moco/tf] test of shape inference for avgpool2d (#3825)

This wll add test of TestShapeTransform AvgPool2D for SAME and VALID padding type

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco exporter] Support AvgPool2D (#3820)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 17 Jun 2019 23:23:18 +0000 (08:23 +0900)]
[loco exporter] Support AvgPool2D (#3820)

This commit supports AvgPool2D loco node.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[angkor] Introduce README.md (#3331)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 17 Jun 2019 22:32:53 +0000 (07:32 +0900)]
[angkor] Introduce README.md (#3331)

* [angkor] Introduce README.md

This commit will add README file for `angkor`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* remove line

* Add words

* Fix headlines

5 years ago[tf2tflite] Ignore name mismatch on testing (#3826)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 17 Jun 2019 11:20:01 +0000 (20:20 +0900)]
[tf2tflite] Ignore name mismatch on testing (#3826)

tfttflite now does not count tensor name mismatch as a test failure.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Test pbtxt extracted from inception_v3 (#3824)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 11:16:35 +0000 (20:16 +0900)]
[moco/tf] Test pbtxt extracted from inception_v3 (#3824)

These test pbtxt files are extracted from inception_v3, containing 4, 7, 10 nodes respectively.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
5 years ago[moco/tf] remove unnecessary comment (#3822)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 10:44:28 +0000 (19:44 +0900)]
[moco/tf] remove unnecessary comment (#3822)

This will remove unnecessary comment about padding

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[nnkit-HDF5] Use index instead of name (#3821)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 17 Jun 2019 10:42:30 +0000 (19:42 +0900)]
[nnkit-HDF5] Use index instead of name (#3821)

Previously, nnkit HDF5 export uses a tensor name when it create a
dataset. With this commit, nnkit HDF5 export uses a tensor index
instead.

This commit also updates nnkit HDF5 import action accordingly.

To maintain backward compatibility, HDF5 export stores the name of
each tensor as an independent attribute. This additional attribute
allows users to detect name mismatch same as before.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] store padding case insensitive (#3819)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 10:26:10 +0000 (19:26 +0900)]
[moco/tf] store padding case insensitive (#3819)

This will change to store padding string as case insensitive as it will be stored once and used many

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] fix padding for avgpool2d (#3816)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 10:23:23 +0000 (19:23 +0900)]
[moco/tf] fix padding for avgpool2d (#3816)

* [moco/tf] fix padding for avgpool2d

This will fox padding calculation formula for AvgPool2D

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* with dilation and log

5 years ago[loco exporter] Refactor Feature Codec export (#3818)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 17 Jun 2019 10:02:41 +0000 (19:02 +0900)]
[loco exporter] Refactor Feature Codec export (#3818)

* [loco exporter] Refactor Feature Codec export

This commit amends detail structure to export FeatureEncode and
FeatureDecode nodes. Mainly, it redefine 'isIdentity()' utility function
and rename 'exportPermutation' into 'exportFeatureTranspose'

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Review fix: rename and notice

5 years ago[nnkit-intf] Update README.md (#3817)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 17 Jun 2019 09:26:16 +0000 (18:26 +0900)]
[nnkit-intf] Update README.md (#3817)

This commit updates nnkit-intf's README.md to be compatible with
Doxygen.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco exporter] General use of checking padding (#3811)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 17 Jun 2019 09:13:06 +0000 (18:13 +0900)]
[loco exporter] General use of checking padding (#3811)

* [loco exporter] General use of checking padding

This commit fixes getOpPadding function. It now supports SAME padding.
Also function signature is generalized to deal with any padding, not just
for MaxPool2d node.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Review fix: careful about unsigned variable

5 years ago[moco/tf] Fix shape transformation for AvgPool2D (#3812)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 07:23:50 +0000 (16:23 +0900)]
[moco/tf] Fix shape transformation for AvgPool2D (#3812)

This will fix shape transformation formula for AvgPool2D with padding type

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] fix maxpool2d shape calculation (#3801)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 07:04:53 +0000 (16:04 +0900)]
[moco/tf] fix maxpool2d shape calculation (#3801)

* [moco/tf] fix maxpool2d shape calculation

This will fix maxpool2d shape calculation as like avgpool2d
- fix also op test model
- add PaddingData annotation

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* fix formula

* fix typo

* rename

5 years ago[tf2tflite] Introduce test infrastructure (#3809)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 17 Jun 2019 06:59:07 +0000 (15:59 +0900)]
[tf2tflite] Introduce test infrastructure (#3809)

* [tf2tflite] Introduce test infrastructure

This commit implements very basic test infrastructure for tf2tflite.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Fix typos

* Update testall

5 years ago[moco/tf] Add comma to logging (#3813)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 17 Jun 2019 05:10:20 +0000 (14:10 +0900)]
[moco/tf] Add comma to logging (#3813)

This will add comma in log string of ShapeTransform of AvgPool2D

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[tftestinfo] isFloatTensor as a const method (#3798)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 17 Jun 2019 03:14:05 +0000 (12:14 +0900)]
[tftestinfo] isFloatTensor as a const method (#3798)

isFloatTensor does not update the content of ParsedTensor.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco exporter] Parse and export graph I/O name (#3803)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 17 Jun 2019 02:02:30 +0000 (11:02 +0900)]
[loco exporter] Parse and export graph I/O name (#3803)

* [loco exporter] Parse and export graph I/O name

Graph input and output names are parsed from loco graph's information
and then exported as name of input and output tensors.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Fix typo

5 years ago[locop] Introduce SymbolTable alias (#3794)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 17 Jun 2019 00:08:47 +0000 (09:08 +0900)]
[locop] Introduce SymbolTable alias (#3794)

This commit introduces SymbolTable alias. This alias make it easy to
refactor code layout.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] File-level comments for tensorflow.test.cpp (#3799)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 09:53:19 +0000 (18:53 +0900)]
[loco] File-level comments for tensorflow.test.cpp (#3799)

This commit adds file-level comments at the top of tensorflow.test.cpp.
These file-level comments are valid for all the tests in this file.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Introduce TensorBiasAdd node (#3783)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 09:14:05 +0000 (18:14 +0900)]
[loco] Introduce TensorBiasAdd node (#3783)

This commit introduces TensorBiasAdd node.

Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco exporter] FeatureCodec identity export (#3795)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 14 Jun 2019 09:09:05 +0000 (18:09 +0900)]
[loco exporter] FeatureCodec identity export (#3795)

Previously FeatureCodec nodes tried to be deleted when they are
identity, but that was not successful. This commit export those as
identity, i.e. CONCATENTATION with one input.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[moco/tf] simplify adding annotation cleaning (#3796)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 08:26:33 +0000 (17:26 +0900)]
[moco/tf] simplify adding annotation cleaning (#3796)

This will rewrite ClearAnnotTransform so that adding cleaning will be bit simple

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] use auto for types in FixShapeTransform (#3793)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 08:18:21 +0000 (17:18 +0900)]
[moco/tf] use auto for types in FixShapeTransform (#3793)

This will fix to use auto for types in FixShapeTransform methods

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] Implement pad fixing for AvgPool2D (#3790)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 08:09:28 +0000 (17:09 +0900)]
[moco/tf] Implement pad fixing for AvgPool2D (#3790)

This will implement pad fixing for AvgPool2D

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] Conv2D test for the 1st Conv2D in inception v3 (#3787)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 07:38:02 +0000 (16:38 +0900)]
[moco/tf] Conv2D test for the 1st Conv2D in inception v3 (#3787)

This adds Conv2D test for the 1st Conv2D in inception v3.
This test failed because the input name is a form of "input:0", which was not correctly processed by moco before.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
5 years ago[tf2tflite] Fix build break (#3786)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 06:41:19 +0000 (15:41 +0900)]
[tf2tflite] Fix build break (#3786)

Recent commit on moco changes the frontend interface, but this change is
not reflected into tf2tflite.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] implement fix shape for avgpool2d (#3777)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 06:29:21 +0000 (15:29 +0900)]
[moco/tf] implement fix shape for avgpool2d (#3777)

This will implement fix_shape for avgpool2d node in fixshape transformation

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] Make moco to use TensorName class (#3781)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 06:12:39 +0000 (15:12 +0900)]
[moco/tf] Make moco to use TensorName class (#3781)

This commit makes 1) moco to use TensorName class 2) renames some naming (e.g., renaming `nodes` to `tensors` if `nodes` actually means `tensors`)

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
5 years ago[tf2tflite] Initial commit (#3779)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 06:00:31 +0000 (15:00 +0900)]
[tf2tflite] Initial commit (#3779)

This commit introduces tf2tflite with minimal implementation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] add AvgPool test model (#3774)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 05:10:22 +0000 (14:10 +0900)]
[moco/tf] add AvgPool test model (#3774)

This will add basic test material for AvgPool import

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Introduce BiasEncode (#3766)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 05:08:23 +0000 (14:08 +0900)]
[loco] Introduce BiasEncode (#3766)

This commit extends loco with BiasEncode node which converts a tensor as
a bias.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] TensorName class to represent the name of tensor (#3776)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 05:06:00 +0000 (14:06 +0900)]
[moco/tf] TensorName class to represent the name of tensor (#3776)

* [moco/tf] TensorName class to represent the name of tensor

This class represents tensor name, which is a form of _nodeName_ ":" _number

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* adds a empty line

5 years ago[moco/tf] Add shapeinfdata feature get,set (#3769)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 04:44:31 +0000 (13:44 +0900)]
[moco/tf] Add shapeinfdata feature get,set (#3769)

* [moco/tf] Add shapeinfdata feature get,set

This will add getter and setter with FeatureShape for ShapeInferenceData

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* remove unused, fix comment

* fix typo

5 years ago[loco exporter] Implement loco exporter v0.1 (#3737)
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Fri, 14 Jun 2019 04:43:16 +0000 (07:43 +0300)]
[loco exporter] Implement loco exporter v0.1 (#3737)

This commit adds minimal implementation that is able to export
a single node network that includes one MaxPool2D op.

Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
Signed-off-by: Efimov Alexander <a.efimov@samsung.com>
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
5 years ago[moco/tf] Introduce AvgPool2D node builder (#3772)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 04:43:05 +0000 (13:43 +0900)]
[moco/tf] Introduce AvgPool2D node builder (#3772)

This will introduce AvgPool2D node builder and related empty methods

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Canonical Node Visitor with default implementation (#3771)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 04:40:06 +0000 (13:40 +0900)]
[loco] Canonical Node Visitor with default implementation (#3771)

* [loco] Canonical Node Visitor with default implementation

With this commit, users are able to implement a node visitor with their
own default implementation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Fix a typo

5 years ago[locop] Show correct opname for canonical dialect (#3768)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 14 Jun 2019 01:59:33 +0000 (10:59 +0900)]
[locop] Show correct opname for canonical dialect (#3768)

With this commit, LinearV1 formatter shows correct opname for
canonical dialect.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Introduce str_toupper (#3770)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 14 Jun 2019 01:43:23 +0000 (10:43 +0900)]
[moco/tf] Introduce str_toupper (#3770)

This will introduce str_toupper() method to convert string to upper case

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] add missing fix_padding (#3764)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 13 Jun 2019 10:19:08 +0000 (19:19 +0900)]
[moco/tf] add missing fix_padding (#3764)

This will add missing fix_padding for Conv2D and FilterEncode

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] clean padding annotation (#3760)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 13 Jun 2019 09:55:39 +0000 (18:55 +0900)]
[moco/tf] clean padding annotation (#3760)

This will add padding data annotation free in transformation

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] basic Conv2D (padding not yet implemented) (#3731)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 13 Jun 2019 09:26:32 +0000 (18:26 +0900)]
[moco/tf] basic Conv2D (padding not yet implemented) (#3731)

* [moco/tf] basic Conv2D (padding not yet implemented)

This adds basic implementation of Conv2D.
Padding and `FixShapeTransform` is not yet implemented.

Note: To run the test, `FixShape()` for Conv2D should be implemented. So I added a minimum implementation in this commit.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* renaming vars

* split one GraphUpdate into two

* assert for padding

* supporting NCHW in FeatureDecode

5 years ago[locop] Initial commit (#3756)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 13 Jun 2019 07:05:53 +0000 (16:05 +0900)]
[locop] Initial commit (#3756)

This commit creates "locop" which provides various pretty-printing
routines for loco IR.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Canonical Nodes (#3759)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 13 Jun 2019 07:00:59 +0000 (16:00 +0900)]
[loco] Canonical Nodes (#3759)

This commit updates the existing nodes to inherit canonical node
interface.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Introduce PaddingData (#3755)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 13 Jun 2019 05:21:17 +0000 (14:21 +0900)]
[moco/tf] Introduce PaddingData (#3755)

This will introduce PaddingData as annotation for converting TF padding to loco padding

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Introduce CanonicalNodeVisitor (#3748)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 13 Jun 2019 03:56:38 +0000 (12:56 +0900)]
[loco] Introduce CanonicalNodeVisitor (#3748)

This commit introduces CanonicalNodeVisitor interface, and adds a related
test.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Introduce empty FixPaddingTransform (#3754)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 13 Jun 2019 03:50:02 +0000 (12:50 +0900)]
[moco/tf] Introduce empty FixPaddingTransform (#3754)

This will introduce empty FixPaddingTransform to convert TF padding to loco padding

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco_exporter] Initial commit (#3757)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 13 Jun 2019 03:49:47 +0000 (12:49 +0900)]
[loco_exporter] Initial commit (#3757)

This commit creates "loco_exporter" project with minimal implementation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Helper function to check equality of two dims (#3727)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 12 Jun 2019 10:52:19 +0000 (19:52 +0900)]
[loco] Helper function to check equality of two dims (#3727)

* [loco] Helper function to check equality of two dims

This adds a helper operator `==` to check equality of two Dimensions.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* modified per comments

5 years ago[moco/tf] make copy_shape_values as template (#3745)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 12 Jun 2019 10:43:53 +0000 (19:43 +0900)]
[moco/tf] make copy_shape_values as template (#3745)

This will change copy_shape_values() to support loco::Node type to reduce duplicate codes

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Remove outdated comments in Nodes.h (#3746)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 12 Jun 2019 09:30:08 +0000 (18:30 +0900)]
[loco] Remove outdated comments in Nodes.h (#3746)

This commit removes several outdated comments in Nodes.h.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Expose verifier and codec through loco.h (#3749)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 12 Jun 2019 09:27:19 +0000 (18:27 +0900)]
[loco] Expose verifier and codec through loco.h (#3749)

This commit updates loco.h to include verifier and permuting codec.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] class storing user-provided compiler arguments (#3734)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 12 Jun 2019 08:51:58 +0000 (17:51 +0900)]
[moco/tf] class storing user-provided compiler arguments (#3734)

* [moco/tf] class storing user-provided compiler arguments

When a user run the compiler with arguments, these arguments are parsed and stored into this class.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* Using default constructor

5 years ago[locomotiv] Conv2D (padding is not yet added) (#3725)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 12 Jun 2019 05:58:56 +0000 (14:58 +0900)]
[locomotiv] Conv2D (padding is not yet added) (#3725)

* [locomotiv] Conv2D (padding is not yet added)

This adds Conv2D implementation of locomotiv. Padding is not implemented yet.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* revised per comments

* Fixed type. Added TODO

* int to uint32_t

5 years ago[moco/tf] a test case for Conv2D (#3728)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 12 Jun 2019 05:33:40 +0000 (14:33 +0900)]
[moco/tf] a test case for Conv2D (#3728)

This adds Conv2D_000 test case for moco/tf.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
5 years ago[loco] Introduce CanonicalNode interface (#3743)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 12 Jun 2019 04:48:13 +0000 (13:48 +0900)]
[loco] Introduce CanonicalNode interface (#3743)

This commit introduces CanonicalNode interface, and rewrites Forward
node as an example.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[locomotiv] Add MaxPool2D test case for padding (#3742)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 12 Jun 2019 03:45:55 +0000 (12:45 +0900)]
[locomotiv] Add MaxPool2D test case for padding (#3742)

This commit test MaxPool2D execution with padding support.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[moco/tf] Clean frontend annotations (#3740)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 12 Jun 2019 03:45:33 +0000 (12:45 +0900)]
[moco/tf] Clean frontend annotations (#3740)

This will clean frontend temporary annotations
- remove test that checks shape information

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Introduce CanonicalOpcode enum class (#3741)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 12 Jun 2019 01:54:34 +0000 (10:54 +0900)]
[loco] Introduce CanonicalOpcode enum class (#3741)

This commit introduces CanonicalOpcode enum class which serves as a
unique tag for each CanonicalNode.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Test case for Concat axis with -1 (#3721)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 11 Jun 2019 10:38:18 +0000 (19:38 +0900)]
[moco/tf] Test case for Concat axis with -1 (#3721)

* [moco/tf] Test case for Concat axis with -1

This will fill the test code for Cocat axis with -1 value

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* check values

* update comment and formula

5 years ago[loco] Extend Node with dialect method (#3730)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 11 Jun 2019 10:04:06 +0000 (19:04 +0900)]
[loco] Extend Node with dialect method (#3730)

This commit extends Node with dialect method as a step toward
multi-dialect support.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Enable concat inference test (#3724)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 11 Jun 2019 10:03:51 +0000 (19:03 +0900)]
[moco/tf] Enable concat inference test (#3724)

This will enable full Concat inference test

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Introduce AvgPool2D (#3720)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 11 Jun 2019 10:02:09 +0000 (19:02 +0900)]
[loco] Introduce AvgPool2D (#3720)

* [loco] Introduce AvgPool2D

This will introduce AvgPool2D for average pool operator.
Attributes are same as that of exising MaxPool2D.

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* add divisor

* rebase

* rebase and add todo

* use full, valid

* uncomment pad

* update note

* use convention

5 years ago[loco] Introduce CanonicalDialect singleton (#3729)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 11 Jun 2019 07:19:20 +0000 (16:19 +0900)]
[loco] Introduce CanonicalDialect singleton (#3729)

This commit introduces CanonicalDialect singleton which serves as an
in-memory unique identifier.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Enable MaxPool2D Pad setter (#3723)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 11 Jun 2019 04:18:16 +0000 (13:18 +0900)]
[loco] Enable MaxPool2D Pad setter (#3723)

This commit enables (temporarily disabled) MaxPool2D Pad setter, and add
a related test.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] add include for stringstream (#3722)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 11 Jun 2019 03:06:45 +0000 (12:06 +0900)]
[moco/tf] add include for stringstream (#3722)

This will add include of sstream that was missed in previous commit

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[locomotiv] MaxPool2D support padding (#3719)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 11 Jun 2019 03:05:06 +0000 (12:05 +0900)]
[locomotiv] MaxPool2D support padding (#3719)

* [locomotiv] MaxPool2D support padding

By this commit, MaxPool2D node execution is able to deal with padding.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Review feedback: local function name

5 years ago[locomotiv] Support FilterEncode execution (#3716)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 11 Jun 2019 02:20:27 +0000 (11:20 +0900)]
[locomotiv] Support FilterEncode execution (#3716)

This commit supports FilterEncode node execution.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[locomotiv] Combine Session test in one file (#3700)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 11 Jun 2019 01:00:14 +0000 (10:00 +0900)]
[locomotiv] Combine Session test in one file (#3700)

Previously Session tests was divided into two files, public and internal
test. This commit combine them into one.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[loco] Introduce replace helper (#3697)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 11 Jun 2019 00:56:50 +0000 (09:56 +0900)]
[loco] Introduce replace helper (#3697)

This commit introduce "replace" helper which facilitate graph rewriting.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Rewrite FixedArityNode as a mix-in (#3699)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 11 Jun 2019 00:55:35 +0000 (09:55 +0900)]
[loco] Rewrite FixedArityNode as a mix-in (#3699)

* [loco] Rewrite FixedArityNode as a mix-in

FixedArityNode now serves as a mix-in rather than a concrete
implementation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Inherit Base

5 years ago[moco/tf] Implement fix_shape for TensorConcat (#3694)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 10 Jun 2019 23:43:47 +0000 (08:43 +0900)]
[moco/tf] Implement fix_shape for TensorConcat (#3694)

* [moco/tf] Implement fix_shape for TensorConcat

This will implement fix_shape in shape inference for TensorConcat node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* add assert and comment

5 years ago[moco/tf] const test with shape inference (#3709)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 10 Jun 2019 10:01:33 +0000 (19:01 +0900)]
[moco/tf] const test with shape inference (#3709)

* [moco/tf] const test with shape inference

This will add test of ShapeInferenceData for ConstGen node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* use name shape

5 years ago[locomotiv] Amend runtime error message for executions (#3712)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 10 Jun 2019 07:41:01 +0000 (16:41 +0900)]
[locomotiv] Amend runtime error message for executions (#3712)

This commit edits runtime error messages for executions. It unifies
message format and style and correct some wrong infos.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[moco/tf] Dump ShapeInferenceData for outputs (#3695)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 10 Jun 2019 07:29:11 +0000 (16:29 +0900)]
[moco/tf] Dump ShapeInferenceData for outputs (#3695)

This will dump ShapeInferenceData for output nodes with moco log library

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[locomotiv] Execution verify and annotate domain info (#3701)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 10 Jun 2019 01:58:04 +0000 (10:58 +0900)]
[locomotiv] Execution verify and annotate domain info (#3701)

This commit makes use of domain annotation tools in node executions.
Now node executions check appropriate domain info preparation, and
annotate appropriate domain info on execution run.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[locomotiv] Node domain annotation for Session (#3682)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 5 Jun 2019 06:39:10 +0000 (15:39 +0900)]
[locomotiv] Node domain annotation for Session (#3682)

Session now annotate node domain on input setting, and delete annotated
domain information on destruction.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
5 years ago[moco/tf] Implement fix_shape for ConstGen (#3692)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 04:32:56 +0000 (13:32 +0900)]
[moco/tf] Implement fix_shape for ConstGen (#3692)

This will implement fix_shape in shape inference for ConstGen node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] Implement fix_shape for Pull (#3693)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 03:54:47 +0000 (12:54 +0900)]
[moco/tf] Implement fix_shape for Pull (#3693)

This will implement fix_shape in shape inference for Pull node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Add Conv2D_Float_000 testcase (#3691)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 5 Jun 2019 03:37:48 +0000 (12:37 +0900)]
[loco] Add Conv2D_Float_000 testcase (#3691)

* [loco] Add Conv2D_Float_000 testcase

This commit adds TensorFlowTest.Conv2D_Float_000 testcase which
checks whether loco is able to encode a simple TensorFlow conv2d
network.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Update per feedback

5 years ago[tfkit] Fix for already packed or unpacked (#3688)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 01:54:49 +0000 (10:54 +0900)]
[tfkit] Fix for already packed or unpacked (#3688)

This will fix tfkit to run pack or unpack for data are already packed or unpakced

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[tfkit] support pack,unpack for int32 type (#3690)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 01:53:38 +0000 (10:53 +0900)]
[tfkit] support pack,unpack for int32 type (#3690)

This will make pack, unpack to support int32 data type

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] Implement fix_shape for copy (#3685)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 01:17:36 +0000 (10:17 +0900)]
[moco/tf] Implement fix_shape for copy (#3685)

This will implement fix_shape() methods for nodes that just will do with copy input shape to output

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Support Padding over Conv2D (#3686)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 5 Jun 2019 00:29:17 +0000 (09:29 +0900)]
[loco] Support Padding over Conv2D (#3686)

This commit extends Conv2D nodes with pad field and its getter.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Introduce Filter Permuting Encoder (#3684)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 5 Jun 2019 00:28:50 +0000 (09:28 +0900)]
[loco] Introduce Filter Permuting Encoder (#3684)

This commit introduces a permuting encoder for filter values.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[moco/tf] Change calling transform run (#3687)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 00:28:41 +0000 (09:28 +0900)]
[moco/tf] Change calling transform run (#3687)

This will change calling transform run by always call and update changed bool value

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[moco/tf] Register two transforms (#3683)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 5 Jun 2019 00:27:54 +0000 (09:27 +0900)]
[moco/tf] Register two transforms (#3683)

This will register and run (currently dummy) two transforms after loading the graph

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
5 years ago[loco] Introduce Pad field and its getter on MaxPool2D (#3679)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 4 Jun 2019 23:40:45 +0000 (08:40 +0900)]
[loco] Introduce Pad field and its getter on MaxPool2D (#3679)

This commit extends MaxPool2D node with a padding field and its getter.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years ago[loco] Implement FilterAxis Permutation (#3675)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 4 Jun 2019 09:30:51 +0000 (18:30 +0900)]
[loco] Implement FilterAxis Permutation (#3675)

* [loco] Implement FilterAxis Permutation

This commit implements Permutation over FilterAxis similarly as that
over FeatureAxis.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Return const reference

* Fix a typo