Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Fri, 9 Nov 2018 14:12:49 +0000 (17:12 +0300)]
[nnc] Add squeeze operation support (#2138)
This commit adds squeeze operation and ShapeInference support for it
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
[nnc] added python model runner. (#2183)
added model runner for caffe, caffe2, onnx and tflite
Signed-off-by: Pavel Fattakhov <p.fattakhov@partner.samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Fri, 9 Nov 2018 11:26:30 +0000 (14:26 +0300)]
[nnc] 'nnmodel' option become a vector (#2157)
Make 'nnmodel' option a vector option, because caffe2 importer consumes two files (init_net.pb and predict_net.pb)
Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 09:29:11 +0000 (18:29 +0900)]
[tflchef] Support AveragePool2D (#2175)
This patch extends tflchef for AveragePool2D operator.
Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 08:27:39 +0000 (17:27 +0900)]
[enco/frontend] Set module inputs and outputs (#2193)
This will set coco Module inputs and outpus from TF lite graph
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 07:45:14 +0000 (16:45 +0900)]
[enco/frontend] Modify get_padding (#2194)
Removed unnecessary assert and added comment for the origin of code.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 9 Nov 2018 07:35:08 +0000 (16:35 +0900)]
[coco] Support ReLU6 for tflite in coco (#2191)
* [coco] Support ReLU6 for tflite in coco
This commit will enable ReLU6 operation for tflite in coco
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* Add comment
* modify comment
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 07:04:12 +0000 (16:04 +0900)]
[enco/frontend] Functions to set coco module IO (#2190)
This will add two methods set_module_inputs and set_module_outputs to
set coco::Module inputs and outputs from TensorContext and TensorBags
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 06:56:31 +0000 (15:56 +0900)]
[enco/frontend] Add function to calculate padding (#2189)
Add function to calculate padding. This will be used by conv2d.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 05:40:40 +0000 (14:40 +0900)]
[enco/frontend] Added OpBuilder, InstrBuilder (#2187)
* [enco/frontend] Added OpBuilder, InstrBuilder
Added OpBuilder, InstrBuilder (copied from enco caffe frontend.)
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* Separate into a new file
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 05:36:51 +0000 (14:36 +0900)]
[enco/frontend] Introduce TensorContext and TensorBags (#2188)
* [enco/frontend] Introduce TensorContext and TensorBags
This will introduce TensorContext to extract and hold informations of
tensors such as shape and name and pre-creates coco::Bag for each tensors
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* apply comments
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 02:19:28 +0000 (11:19 +0900)]
[enco/frontend] GraphBuilder for tflite op and factory (#2171)
This commit introduces GraphBuilder, and its child classes and factory.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 9 Nov 2018 02:15:18 +0000 (11:15 +0900)]
[ANN] Introduce ReLU6 ops (#2176)
* [ANN] Introduce ReLU6 ops
This commit will introduce ReLU6 ops for ANN and execute ReLU6 in ANN
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* align column
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 9 Nov 2018 02:14:56 +0000 (11:14 +0900)]
[tflchef] Support ReLU6 operation (#2179)
* [tflchef] Support ReLU6 operation
This commit will enable ReLU6 operation in tflchef
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* resize the channel size
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 9 Nov 2018 02:07:50 +0000 (11:07 +0900)]
[enco.caffe] Remove redundant layout allocations (#2180)
There are many redundant FeatureLayout allocations in the current caffe
frontend; frontend currently allocates a FeatureObject with deprecated
FeatureObject create method which internally allocates Generic FeatureLayout),
and then updates it with a valid FeatureLayout.
This commit aims to eliminate such redundant layout allocations.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 01:19:18 +0000 (10:19 +0900)]
[enco/frontend] Introduce as_tensor_shape (#2181)
This will introduce as_tensor_shape method to convert flatbuffers::Vector
as nncc::core::ADT::tensor::Shape that will be used for reading tensor shape
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 9 Nov 2018 01:12:52 +0000 (10:12 +0900)]
[tflchef] Bugfix of wrong variable in CMakeLists.txt (#2185)
Originally, `RECIPE_PREFIX` is used for indicating the prefix of each test
However until now, `PREFIX` is used for that
This commit will fix it as correct variable name
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 9 Nov 2018 00:55:59 +0000 (09:55 +0900)]
[enco] fix as_index_vector usage of Vector methods (#2182)
This will fix to use correct flatbuffers::Vector method and type
- Length() is deprecated, should use size()
- size return type is uint32_t not int
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
[nnc] Refactor caffe importer to support layers with multiple outputs (#2154)
- add type alias IODescriptorVector;
- make layer conversion methods accept IODescriptorVector and return IODescriptorVector;
- add support for InputLayer with several inputs and single shape;
- refactor according to coding style.
Signed-off-by: Sergei Barannikov s.barannikov@samsung.com
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Thu, 8 Nov 2018 10:18:02 +0000 (19:18 +0900)]
[enco/frontend] Introduce as_index_vector (#2177)
* [enco/frontend] Introduce as_index_vector
This will introduce as_index_vector method to convert
flatbuffers::Vector to IndexVector as
std::vector<int32_t> that will be used to for reading
index values of operators and operands.
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* add a blank line
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 8 Nov 2018 07:47:06 +0000 (16:47 +0900)]
[enco] Introduce enco_dump_all_ops debugging helper (#2170)
This commit introduces enco_dump_all_ops debugging helper which dumps
all the allocated Ops in a given module.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 8 Nov 2018 01:19:34 +0000 (10:19 +0900)]
[enco] Introduce enco_dump_op debugging helper (#2152)
This commit introduces enco_dump_op helper function to allows users
to inspect the details of coco::Op value from a debugger such as GDB.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 8 Nov 2018 00:48:32 +0000 (09:48 +0900)]
Use PRE_BUILD hook in FlatBuffers_Target (#2161)
It turns out that Ninja failed to propely manage the dependencies of
generated files of a STATIC library.
To address this Ninja build issue, this commit revises FlatBuffers_Target
to insert explicitly dependencies between target and generated files via
using PRE_BUILD hook.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 7 Nov 2018 07:18:09 +0000 (16:18 +0900)]
[enco.caffe] Simplify code using padding helpers (#2151)
This commit simplifies ConvolutionSpec and PoolingSpec code using
build_raw_padding and build_spatial_padding helpers.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 7 Nov 2018 07:17:56 +0000 (16:17 +0900)]
[tflchef] Embed TensorFlow Lite schema file (#2150)
This commit embeds TensorFlow Lite schema file in tflchef in order to
reduce external dependencies.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 7 Nov 2018 07:17:31 +0000 (16:17 +0900)]
[pp] Build as Position Independent Code (#2149)
pp is currently built as a non-PIC static library, and thus it is
impossible to link pp against another shared library.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 7 Nov 2018 05:15:58 +0000 (14:15 +0900)]
[tflchef] Remove dependency to stdex (#2146)
Getting rid of `stdex::make_unique` to minimize dependency.
Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 7 Nov 2018 01:39:55 +0000 (10:39 +0900)]
[enco.caffe] Introduce Padding-related utilities (#2139)
This commit introduces the following helper functions for Padding
variables:
- build_raw_padding
- build_spatial_padding
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 7 Nov 2018 01:19:49 +0000 (10:19 +0900)]
[tflchef] Fix typo in comment (#2145)
I've spotted several typos during reading tflchef code.
This patch fixes them.
Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 7 Nov 2018 00:13:24 +0000 (09:13 +0900)]
[tflchef] Introduce GaussianFloat32DataChef (#2137)
This commit introduces GaussianFloat32DataChef (along with its factory class) and
registers it to the global DataChef registry.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 09:31:01 +0000 (18:31 +0900)]
[enco.caffe] Introduce RawPadding and SpatialPadding (#2134)
This commit introduces RawPadding/SpatialPadding classes as the first
step toward unifying Padding-related routines.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 08:32:12 +0000 (17:32 +0900)]
Do NOT propagate sources (#2132)
This commit revises FlatBuffers_Target not to propagate sources (to
address issue related with the scope of generated files).
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 08:31:17 +0000 (17:31 +0900)]
[tflchef] Extract "to_number" helper (#2131)
This commit extracts "to_number" helper (currently in Data/Constant.h)
to a seperate module.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 06:09:39 +0000 (15:09 +0900)]
[enco.caffe] Construct ConcatF from Frontend (#2126)
This commit revises caffe frontend to construct ConcatF instead of
directly lowering it as a sequence of Shuffle.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 00:46:27 +0000 (09:46 +0900)]
[coco] Remove KernelObject create method with shape (#2125)
This commit removes deprecated KernelObject "create" method with shape
from ObjectManager.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 00:46:17 +0000 (09:46 +0900)]
[enco.tflite] Show the list of failed tests (#2121)
This commit revises "runall" to show the list of failed tests.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 00:28:41 +0000 (09:28 +0900)]
Declare STATIC library in FlatBuffers_Target (#2123)
This commit revises FlatBuffers_Target function to declare STATIC
library instead of INTERFACE library. This change addresses the scope
issue related with generated files.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 6 Nov 2018 00:28:25 +0000 (09:28 +0900)]
[enco.caffe] Remove morph::caffe prefix (#2124)
This commit introduces 'using namespace morph::caffe' to simplify caffe
frontend implementation.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 5 Nov 2018 23:33:34 +0000 (08:33 +0900)]
[enco] Introduce ConcatLowering pass (#2109)
This commit implements ConcatLowering pass in enco backend, which lowers
ConcatF as a sequence of Shuffle.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 5 Nov 2018 07:51:47 +0000 (16:51 +0900)]
[enco.tflite] Write test logs to a file (#2118)
This commit revises enco tflite test framework to write test logs to a
file (to make it easy to analyze test failures).
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 5 Nov 2018 04:51:00 +0000 (13:51 +0900)]
Add tflchef to contrib precedence list (#2114)
* Add tflchef to contrib precedence list
This will add tflchef to precedence as enco will use tflchef as tflite generator
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* edit comments
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 5 Nov 2018 01:05:50 +0000 (10:05 +0900)]
[morph] Add TensorFlow Lite support (#2110)
* [morph] Add TensorFlow Lite support
This commit extends morph with tensorflow lite support.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Add missing license notice
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 5 Nov 2018 01:05:36 +0000 (10:05 +0900)]
[coco] Rename 'Axe' as 'Axis' (#2113)
This commit renames 'Axe' enum class of 'ConcatF' operation as 'Axis'
to make it easy to understand.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Sun, 4 Nov 2018 23:15:30 +0000 (08:15 +0900)]
Remove logging code inside FlatBuffers_Target (#2106)
The committed implementation of FlatBuffers_Target includes unnecessary
logging code by mistake.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
[nnc][utils] Add generator of random tflite models. (#2111)
Currently the generator supports 2 operations
and generate only constant tensor's shapes.
Signed-off-by: i-kulagin <i.kulagin@samsung.com>
Enable ONNX frontend compilation (#2082)
Special --onnx option added.
ONNX operator types aadded.
CMake build system updated to add ONNX related stuff.
Signed-off-by: Andrew V. Tischenko <a.tischenko@partner.samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 2 Nov 2018 06:59:24 +0000 (15:59 +0900)]
[enco.caffe] Remove LoadOpBuilder (#2105)
This commit removes LoadOpBuilder, and rewrites relavent code using
OpBuilder.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 2 Nov 2018 06:52:56 +0000 (15:52 +0900)]
[enco] Write caffe test log to a file (#2103)
This commit revises 'runall' script (in enco caffe test framework) to
log all the output (stdout/stderr) to a file to make it easy to
diagnose test failure.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 2 Nov 2018 04:17:27 +0000 (13:17 +0900)]
[encodump] Include "functional" (#2102)
std::function is referenced, but functional is not included.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 2 Nov 2018 00:06:22 +0000 (09:06 +0900)]
[enco] Introduce enco test framework for tflite (#2079)
* [enco] Introduce enco test framework for tflite
Until now, enco test framework for tflite was not supported
This commit will enable enco test for tflite using empty recipe file
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* Modify Target Check
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 1 Nov 2018 23:38:48 +0000 (08:38 +0900)]
[enco] Use FlatBuffers_Target (instead of FlatBuffers_Generate) (#2097)
This commit replaces FlatBuffers_Generate with FlatBuffers_Target.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
[nnc] ACL backend elementwise, deconvolution and elu (#2100)
Elementwise, Deconvolution, and BiasAdd implementations and marking Elu as not supported yet.
The operations listed in the title done. And now, formally, all the operations are handled. Some of them marked as unsupported.
Signed-off-by: Timur Ablyazimov t.ablyazimov@samsung.com
[nnc] Rename files with ModelIR operation classes according to coding style (#2099)
* Rename files containing classes for ModelIR operations (include/core/modelIR/operations/)
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Thu, 1 Nov 2018 08:40:38 +0000 (17:40 +0900)]
[encodump] Prints bags, objects, ops (#2077)
[encodump] Prints bags, objects, ops
`encodump` prints bags, objects, ops. Plus minor enhancements were made.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 1 Nov 2018 08:37:39 +0000 (17:37 +0900)]
[enco] Update comment of caffe enco test option name (#2090)
This commit will update comment of `ENCO_CAFFE_TEST` option
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 1 Nov 2018 08:37:13 +0000 (17:37 +0900)]
Introduce FlatBuffers_Target function (#2085)
This commit introduces FlatBuffers_Target function which directly creates
a target.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 1 Nov 2018 08:15:35 +0000 (17:15 +0900)]
[enco] Move binder.cpp for re-using in enco-test (#2086)
If enco-test for TFLite is added, it will share exactly same binder.cpp
Therefore it is better to move binder.cpp to `enco/test` folder, which makes sharing much easier
This commit will move binder.cpp and modify CmakeLists.txt of caffe to apply the change
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 10:04:20 +0000 (19:04 +0900)]
[enco] TensorFlow Lite frontend (v0.1) (#2076)
With this commit, enco tflite frontend is now able to construct
coco::Module and coco::Data from "empty" tflite model.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 09:46:16 +0000 (18:46 +0900)]
[enco] Introduce instr_sequence helper (#2061)
This commit introduces instr_sequence helper function which returns a
instruction sequence (in execution order).
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 31 Oct 2018 09:10:19 +0000 (18:10 +0900)]
[encodump] adds initial visit function for ConcatF (#2062)
Created visit function for new op `ConcatF`. Without this, calling `visit(concatF)` throws an runtime exception.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 07:47:15 +0000 (16:47 +0900)]
[coco] Op::Visitor as non-pure virtual class (#2060)
This commit renames Op::Visitor as Op::IVisitor, and introduce the
default implementation for each visit method.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 31 Oct 2018 07:40:20 +0000 (16:40 +0900)]
[enco] Update usage message (#2058)
Usage message for prefix was omitted
This commit will fix it
Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 07:40:10 +0000 (16:40 +0900)]
[enco.tflite] RawModel-based Frontend template (#2054)
This commit revises (tflite) Frontend class to take a RawModel as an
argument of its constructor, and implements the basic template of "load"
method.
Note that the current implementation does not implement "make_frontend"
yet.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 07:36:44 +0000 (16:36 +0900)]
[enco] Show failed tests (#2057)
This commit revises "runall" to show the list of failed tests.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 07:35:02 +0000 (16:35 +0900)]
[coco] Introduce depth/height/width to FeatureLayout (#2056)
For consistency, this commit introduces depth/height/width method to
FeatureLayout class.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 31 Oct 2018 07:34:29 +0000 (16:34 +0900)]
[enco] tool to dump IR (initial version) (#2034)
This tool prints IR info of caffe frontend.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 03:15:56 +0000 (12:15 +0900)]
[enco] Fix Ninja build error (#2053)
This commit updates target include directory to address Ninja build
error.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 01:46:45 +0000 (10:46 +0900)]
[coco] Fix a bug in BC FeatureLayout test (#2049)
The current test uses invalid 'batch' value.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 01:46:20 +0000 (10:46 +0900)]
[coco] Check Op-to-Instr link on destruction (#2048)
This commit inserts assert over _step variable in destructor in order to
guarantees that Op is not referred by any Instr.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 31 Oct 2018 01:39:39 +0000 (10:39 +0900)]
[doc] Add format checker section into project_guide.md (#2040)
Add how to enable format checker for incubating project.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 01:05:55 +0000 (10:05 +0900)]
[coco] Rename GenericKernelLayout class (#2026)
This commit renames GenericKernelLayout class as Generic class under
KernelLayouts namespace, which makes it consistent with other layout
classes.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 31 Oct 2018 00:44:08 +0000 (09:44 +0900)]
[enco.tflite] Add missing Frontend constructor (#2045)
This commit implements missing (tflite) Frontend constructor, and adds
related tests.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 30 Oct 2018 17:50:14 +0000 (20:50 +0300)]
[nnc] Do not use non-interface includes in importers (#2052)
- do not use non-interface includes in importers
- do not build frontend unittest if frontend disabled
Signed-off-by: Ivan Vagin ivan.vagin@samsung.com
[nnc] Rename op creation methods to match framework layer names (#2051)
* Rename CaffeOpCreator and TFLiteOpCreator methods to match source framework layer names, not the MIR operation names
Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
Initial top-level ONNX support (#1987)
CMake changes were done to allow ONNX downloading and onnx.proto compiling.
Signed-off-by: Andrew V. Tischenko a.tischenko@partner.samsung.com
[nnc] Remove Model IR after all passes (#2036)
* add `cleanup` method in `Pass` class
* remove redundant frontend classes
* free all pass data in `PassManager` destructor
* replace `OpCreator` with CaffeOpCreator/TFLiteOpCreator to avoid violation of ODR
* free resources in unit tests
Signed-off-by: Roman Rusyaev <r.rusyaev@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Oct 2018 06:01:45 +0000 (15:01 +0900)]
Configure contrib build flag per each configuration (#2041)
BUILD_CONTRIB_<TAG> value is fixed at the first configure run and never
updated with the current implementation.
With this commit, BUILD_CONTRIB_<TAG> value is updated properly per each
configure run.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Oct 2018 04:27:46 +0000 (13:27 +0900)]
[tflchef] Introduce test framework (#2032)
* [tflchef] Introduce test framework
This commit introduces a simple test framework for tflchef.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Update comment
* Rename testcase to make it easy to understand
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Oct 2018 04:23:08 +0000 (13:23 +0900)]
[coco] Introduce ConcatF operation (#2029)
This commit introduces ConcatF operation which concatenates two feature
objects along a specified axis.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Oct 2018 04:21:39 +0000 (13:21 +0900)]
[enco] Allow users to disable caffe test (#2037)
With this commit, users are now able to disable caffe test via CMake configuration.
Note that this commit does not change the default behavior.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Oct 2018 04:06:21 +0000 (13:06 +0900)]
[enco] Support Scale with bias term (#2030)
This commit extends enco caffe frontend to support Scale layer with bias term,
and introduces related caffe test.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
Update project documents for DR2 (#1640)
This commit add Unit test report and updates SDD, STD, DLD
for DR2 milestone
Signed-off-by: Sergey Vostokov <s.vostokov@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 29 Oct 2018 07:33:29 +0000 (16:33 +0900)]
[enco] Support simple Scale layer (#2027)
This commit extends caffe frontend to support simple Scale layer (a
Scale layer without bias term).
This commit also includes a related test case.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 29 Oct 2018 01:08:22 +0000 (10:08 +0900)]
[enco] Introduce InstrBuilder (#2015)
This commit introduces InstrBuilder and simplifies the implementation of
caffe frontend.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
[nnc] Acl backend scale batchnorm dropout (#1954)
Scale and DropOut operations support in ACL backend.
Signed-off-by: Timur Ablyazimov <t.ablyazimov@samsung.com>
[nnc] Style transfer importer (#1892)
This PR adds Style Transfer operations support to caffe Importer.
The operations are:
- Deconvolution
- ELU
- TanH
- Elementwise
Signed-off-by: Andrei Shedko a.shedko@partner.samsung.com
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 26 Oct 2018 09:24:43 +0000 (18:24 +0900)]
[enco] Add tflite schema generation (#2014)
* [enco] Add tflite schema generation
This will add tflite schema generation
- add schema file with meta information file
- change cmake for header generation
- generated header is included in Entry.cpp for testing integrity
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* apply comments
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 26 Oct 2018 07:33:01 +0000 (16:33 +0900)]
[enco] Introduce OpBuilder (#2012)
This commit introduces OpBuilder class, and simplifies the
implementation of caffe frontend using OpBuilder.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 25 Oct 2018 17:51:17 +0000 (20:51 +0300)]
[nnc] Impement node replacement facility in Graph class (#1951)
Provides replaceNode which is used to replace a node in graph preserving all edges
Provides replaceNodeWithInput and replaceInputNodes methods, used to replace any node with input node
Provides replaceOutputNodes, used to replace graph output nodes with new list
Adds unit tests for all implementations
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
[nnc] Acl backend activations and reshape (#1952)
ReLU, Capped ReLU, Tanh and Reshape support.
Signed-off-by: Timur Ablyazimov <t.ablyazimov@samsung.com>
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 25 Oct 2018 17:42:47 +0000 (20:42 +0300)]
[nnc] Resolve bug with custom opcodes unsupported in tflite (#2007)
Resolves bug #1991 with unsupported custom opcodes
Signed-off-by: Vladimir Plazun <v.plazun@partner.samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 25 Oct 2018 08:26:05 +0000 (17:26 +0900)]
[coco] Remove unused PtrSlot.h (#1985)
This commit removes unused PtrSlot.h and related tests.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 25 Oct 2018 08:25:34 +0000 (17:25 +0900)]
[enco] Introduce 'build_load' helper (#1983)
* [enco] Introduce 'build_load' helper
This commit introduces 'build_load' helper, and simplifies the
implementation of caffe frontend by using this helper.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Remove unnecessary object update
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 25 Oct 2018 00:36:38 +0000 (09:36 +0900)]
[coco] Remove reads/updates from Instr (#1968)
With this commit, there is no need to implement reads/updates method for
each Instr, which makes it easy to introduce a new instruction.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 24 Oct 2018 17:54:33 +0000 (20:54 +0300)]
Refactored tflite frontend and fixed coding style (#1885)
Refactored tflite frontend and fixed coding style
Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Oct 2018 10:35:52 +0000 (19:35 +0900)]
[enco] Remove unused reads/updates (#1963)
This commit removes unused reads/updates helper functions.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Oct 2018 08:19:31 +0000 (17:19 +0900)]
[coco] Introduce LLVM-like Instr casting operators (#1962)
This commit introduces the following LLVM-like casting operators for Instr class:
- isa<T>
- cast<T>
- safe_cast<T>
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Oct 2018 08:18:26 +0000 (17:18 +0900)]
[enco] Identify ANN input/output with ANN bags method (#1960)
* [enco] Identify ANN input/output with ANN bags method
This commit refines the implementation of ANNModuleBuilder to use bags
method in ANNBinder instead of reads/updates method.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Remove unused 'block' variable
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Oct 2018 02:12:47 +0000 (11:12 +0900)]
[enco] Extend ANNBinder with bags method (#1959)
This commit extends ANNBinder with bags method which enumerates all the
bags that the current ANN subnet accesses during evaluation.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Oct 2018 02:09:46 +0000 (11:09 +0900)]
[enco] Take "const Bag *" as an input (#1958)
The current implementation of readers/updaters takes "Bag *" as an
input, but these functions do not update Bag itself.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>