박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 18 Jul 2019 01:35:30 +0000 (10:35 +0900)]
[loco] Shaped Graph Input/Output (#4325)
This commit extends GraphInput & GraphOutput with the field and methods
related with shape.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 18 Jul 2019 01:33:00 +0000 (10:33 +0900)]
Introduce EigenSource package for TensorFlow 1.12. (#4328)
This commit introduces "EigenSource-
fd6845384b86" for TensorFlow 1.12.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 18 Jul 2019 01:30:56 +0000 (10:30 +0900)]
[loco] Introduce DepthwiseConv2D (#4322)
This commit introduce DepthwiseConv2D which represents depthwise
2D convolutional operation.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 18 Jul 2019 01:24:47 +0000 (10:24 +0900)]
[moco/tf] fix TFNodeSummaryBuilder TFRsqrt (#4334)
This will add TFRsqrt for TFNodeSummaryBuilder that was recently added
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 18 Jul 2019 00:49:30 +0000 (09:49 +0900)]
[moco/tf] TF Formatted Graph support (#4310)
* [moco/tf] TF Formatted Graph support
This will introduce TF node summary builder and related classes to dump valid TF dialect names in graph dump
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* rename class
* remove not ready node
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 18 Jul 2019 00:04:32 +0000 (09:04 +0900)]
[moco/tf] Log helper (#4323)
* [moco/tf] Log helper
This will add helper functions to dump Feature or Filter to stream
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* use loco namespace
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 09:58:02 +0000 (18:58 +0900)]
[moco/tf] Remove Knob in Const Importer unittest (#4316)
This commit removes all the use of "Knob" in Const Importer unittest.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 17 Jul 2019 07:56:29 +0000 (16:56 +0900)]
[moco/tf] more logs in fix shape for const (#4315)
* [moco/tf] more logs in fix shape for const
This will log Const node shape information in shape inference
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* use helper
* include header
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 07:49:38 +0000 (16:49 +0900)]
[loco] Introduce CanonicalNodeDef interface (#4302)
This commit introduces CanonicalNodeDef interfaces which takes multiple
Mixins all at once.
Compared with existing CanonicalNodeImpl, CanonicalNodeDef emits uniform
declarations that begin with "CanonicalNodeDef<OPCODE, ...>", which makes
it easy to read the code.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 17 Jul 2019 04:58:55 +0000 (13:58 +0900)]
[moco/tf] Introduce TFRsqrt IR (#4317)
This will introduce TFRsqrt IR and related codes
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 04:06:48 +0000 (13:06 +0900)]
[moco/tf] Remove Knob in Conv2D importer test (#4303)
This commit removes Knob in Conv2D importer test. Each unittest run
now tests both loco.TF and loco.Canonical importers at once.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 03:49:55 +0000 (12:49 +0900)]
[exo/tflite] Use Visitor in Operation Exporter (#4300)
This commit revises the implementation of Operation Exporter using loco
visitor interface.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 03:48:46 +0000 (12:48 +0900)]
[exo/tflite] Visitor-based Type Inference (#4296)
* [exo/tflite] Visitor-based Type Inference
This commit revises the implemenation of TypeInference module using loco
visitor interface.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Fix subobject-linkage warning
* Explain why there is type context update
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 03:11:00 +0000 (12:11 +0900)]
[exo/tflite] Visitor-based Shape Inference (#4299)
* [exo/tflite] Visitor-based Shape Inference
This commit revises the implementation of ShapeInference using loco
visitor interface.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Fix subobject-linkage warning
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 17 Jul 2019 01:45:32 +0000 (10:45 +0900)]
[moco/tf] ResolveFBN for only active nodes (#4311)
This will change ResolveFusedBatchNorm transformation to iterate only active nodes
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 17 Jul 2019 01:44:52 +0000 (10:44 +0900)]
[moco/tf] Canonicalize Const (#4312)
This will add canonicalization for Const node
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 01:36:37 +0000 (10:36 +0900)]
Relocate CMake extensions (#4308)
This commit relocates top-level cmake directory into infra/nncc, and
updates relevant top-level CMakeLists.txt
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 01:35:25 +0000 (10:35 +0900)]
Relocate nncc commands (#4307)
All the nncc commands are now in infra/nncc/command.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 17 Jul 2019 00:00:30 +0000 (09:00 +0900)]
Relocate nncc configurations (#4248)
This commit introduces "infra/nncc/config" directory, and move all the
.configuration files into this directory.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 17 Jul 2019 00:00:14 +0000 (09:00 +0900)]
[moco-tf] Setting basic data of constgen in test (#4304)
After creating `constgen` in `optimizer.test.cpp`, `dtype` and `size` are now set to proper values.
Some transforms (which is being written) needs dtype and values and this commit enables these transforms to run OK.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 16 Jul 2019 10:34:23 +0000 (19:34 +0900)]
[res] TF_SMALL_NET_0010 (#4301)
This will add TF_SMALL_NET_0010 test that has "BiasAdd" + "Add" network
This is to test fusion of Add into BiasAdd
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 16 Jul 2019 07:24:31 +0000 (16:24 +0900)]
[locomotiv] Prevent input/output index overflow (#4286)
This commit make assertion on input and output index to be less than
their size.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 16 Jul 2019 07:23:50 +0000 (16:23 +0900)]
[res/test] TF_SMALL_NET_0009 (#4287)
This network is Conv2D + BiasAdd + Add to check fusion of BiasAdd + Add to BiasAdd
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 16 Jul 2019 07:06:07 +0000 (16:06 +0900)]
[locoex/customop] Adding COpCall node (#4290)
COpCall node (previously merged as Call node) is added with some methods.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 16 Jul 2019 06:24:30 +0000 (15:24 +0900)]
[moco/tf] Test BiasAdd Importer without Knob (#4265)
This commit removes Knob from BiasAdd test implementation.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 16 Jul 2019 06:17:57 +0000 (15:17 +0900)]
[moco/tf] Use TFConst for Resolve FBN (#4291)
This will change ResolveFusedBatchNorm transformation to use TFConst instead of ConstGen as an inputs
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 16 Jul 2019 04:22:32 +0000 (13:22 +0900)]
[locoex/customop] Change name of file (#4288)
Previous `AttrTypes.h` is now renamed to `COpAttrTypes.h`
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 16 Jul 2019 02:57:21 +0000 (11:57 +0900)]
[loco] Permuting encoder for DepthwiseFilter (#4284)
This commit introduces permuting encoder for depthwise filter domain.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 16 Jul 2019 01:28:01 +0000 (10:28 +0900)]
[moco/tf] Skip fix_shape for add when disconnected (#4282)
This will skip fix_shape for TFAdd node when it's disconnected from some transformation
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 16 Jul 2019 00:25:16 +0000 (09:25 +0900)]
[loco] Permutation for DepthwiseFilter (#4280)
This commit introduces permutation between DepthwiseFilter and Tensor.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 15 Jul 2019 10:40:17 +0000 (19:40 +0900)]
[loco] Introduce DepthwiseFilterAxis (#4278)
This commit introduces DepthwiseFilterAxis to describe each axis of
depthwise convolution filter.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 15 Jul 2019 10:40:08 +0000 (19:40 +0900)]
[loco] Introduce DepthwiseFilter Domain (#4277)
This commit introduces new DepthwiseFilter Domain. Also comment to
clarify meaning between Filter and DepthwiseFilter is added.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 10:21:30 +0000 (19:21 +0900)]
[loco] Extract SymbolTable class (#4276)
This commit extracts SymbolTable class into a separate file (to simplify
FormattedGraph.h).
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 10:01:26 +0000 (19:01 +0900)]
[moco/tf] Revise Const unit test (#4271)
This will revise Const unit test to test TFConst with a knob and test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 10:00:02 +0000 (19:00 +0900)]
[moco/tf] Implement fix_shape for TFConst (#4274)
This will implement fix_shape() for TFConst node
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 09:59:28 +0000 (18:59 +0900)]
[moco/tf] Revise FusedBatchNorm unit test (#4272)
This will revise FusedBatchNorm unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 15 Jul 2019 09:53:48 +0000 (18:53 +0900)]
[loco] Introduce DepthwiseFilterEncode (#4273)
This commit introduces DepthwiseFilterEncode loco canonical dialect
node. This node encodes tensor from input node into depthwise filter.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 09:07:18 +0000 (18:07 +0900)]
Extract StampTools module (#4268)
This commit extracts StampTools module from ExternalSourceTools.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 15 Jul 2019 09:03:18 +0000 (18:03 +0900)]
[loco] Introduce DepthwiseFilterEncoder (#4270)
DepthwiseFilterEncoder describes how tensor should be encoded into
depthwise convolutional filter.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 08:11:17 +0000 (17:11 +0900)]
[mocotest/tf] Enable Relu6_000 test (#4266)
Let's check whether moco.tf frontend works well for Relu6_000 via
locomotiv.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 08:09:36 +0000 (17:09 +0900)]
[moco/tf] Fix ConstCanonicalizer (#4269)
This will fix ConstCanonicalizer to allocate const data buffer
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 08:09:13 +0000 (17:09 +0900)]
[moco/tf] Revise Conv2D unit test (#4267)
This will revise Conv2D unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 08:08:40 +0000 (17:08 +0900)]
[moco/tf] Revise Concat unit test (#4244)
* [moco/tf] Revise Concat unit test
This will revise Concat unit test test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* use find_first_node_bytype
* apply comments
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 15 Jul 2019 07:49:16 +0000 (16:49 +0900)]
[loco] Introduce DepthwiseFilterIndex (#4263)
DepthwiseFilterIndex represents specific index in DepthwiseFilterShape.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 07:41:28 +0000 (16:41 +0900)]
[moco/tf] Import as TFConst (#4261)
This will introduce Knbo for import as TFConst and the changes to import Const as TFConst node
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 07:40:54 +0000 (16:40 +0900)]
[moco/tf] Canonicalize Const (#4252)
* [moco/tf] Canonicalize Const
This will introdude ConstCanonicalize to canonicalize TFConst to ConstGen
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* fix comment
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 07:05:44 +0000 (16:05 +0900)]
[moco/tf] Check canonicalization (#4257)
* [moco/tf] Check canonicalization
This will add a check routine if canonicalization has really done its task
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* extract helper function
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 06:32:50 +0000 (15:32 +0900)]
[locoex/customop] make ctor of COpAttrData protected (#4260)
This commit makes ctor of COpAttrData protected since the ctor is called only by children.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 06:32:13 +0000 (15:32 +0900)]
[moco/tf] Chainable GraphBuilderRegistry (#4258)
This commit allows users to build a GraphBuilderRegistry chain.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 06:32:02 +0000 (15:32 +0900)]
[moco/tf] Importer with User-Defined GraphBuilderSource (#4259)
This commit introduces a new Importer constructor which allows users to
inject their-own GraphBuilderSource to Importer.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 05:29:55 +0000 (14:29 +0900)]
[locoex/customop] Introducing COpNode class (#4253)
* [locoex/customop] Introducing COpNode class
This commit adds COpNode class for locoex custom op.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* remove COpNodeImpl
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 05:07:04 +0000 (14:07 +0900)]
Introduce tfgraph-xform (#4251)
* Introduce tfgraph-xform
Let's build TensorFlow "transform_graph" without Bazel!
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Update comments
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 05:05:48 +0000 (14:05 +0900)]
Add TensorFlow Relu6 test (#4256)
This commit adds Relu6_000 TensorFlow testcase.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 04:47:11 +0000 (13:47 +0900)]
[exo/tflite] Extract ShapeInference (#4255)
This commit introduces ShapeInference.h and ShapeInference.cpp and
extracts all the declarations and implementations related with
ShapeInference from TypeInference.h & TypeInference.cpp.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 15 Jul 2019 04:43:37 +0000 (13:43 +0900)]
[loco] Introduce DepthwiseFilterShape (#4250)
DepthwiseFilterShape represents shape for filter used by Depthwise
2D Convolution.
Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:58:09 +0000 (10:58 +0900)]
[moco/tf] Revise AvgPool2D unit test (#4242)
This will revise AvgPool2D unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 01:47:55 +0000 (10:47 +0900)]
Add execution permission to make_directories (#4249)
The current make_directories in TensorFlowProtoText package has no
execution permission.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:44:04 +0000 (10:44 +0900)]
[moco/tf] Introduce TFConst (#4240)
This will introduce TFConst IR and related minimum codes
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:43:42 +0000 (10:43 +0900)]
[moco/tf] Fix PadData and PaddingData header (#4212)
This will fix comments of PadData and PaddingData where brief is too long
and also some notes about difference in between to lower confusion.
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:36:41 +0000 (10:36 +0900)]
[moco/tf] Revise ReLU6 unit test (#4247)
This will revise ReLU6 unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:36:22 +0000 (10:36 +0900)]
[moco/tf] Revise ReLU unit test (#4246)
This will revise ReLU unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:36:04 +0000 (10:36 +0900)]
[moco/tf] Revise MaxPool2D unit test (#4245)
This will revise MaxPool2D unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:35:23 +0000 (10:35 +0900)]
[moco/tf] Revise BiasAdd unit test (#4243)
This will revise BiasAdd unit test to test only related properties
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:22:10 +0000 (10:22 +0900)]
[locoex/customop] Attr types and data structure (#4203)
* [locoex/customop] Attr types and data structure
This file contains attribute types and structs to hold the value of attr types.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* renaming: attrType() -> type()
* removed customop dir and namespace
* remvoe attrTypeTraits. This will be committed later
* Rename class name
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:21:53 +0000 (10:21 +0900)]
[locoex/customop] Dialect for Customop (#4209)
* [locoex/customop] Dialect for Customop
This adds Dialects declaration for custom op.
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* remove 'customop' dir and milestone. change class name.
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 01:20:20 +0000 (10:20 +0900)]
[exo/tflite] Do NOT export Identity (#4237)
This commit revises TensorAllocation to allocate the same operand
to the input and output of "identity"-like operations, and revises
OperationExporter not to export such operations.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 01:18:48 +0000 (10:18 +0900)]
Introduce TensorFlowProtoText 1.12 package (#4230)
* Introduce TensorFlowProtoText 1.12 package
This commit introduces TensorFlowProtoText 1.12 package, which provies
"proto_text" executable (with a different name) required for TensorFlow
build.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Be compatibile with CMake 3.1
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 01:09:24 +0000 (10:09 +0900)]
Introduce and use import function for configuraiton loading (#4206)
This commit revises nncc & command scripts not to depend on
"NNCC_SCRIPT_PATH" variable.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 01:05:56 +0000 (10:05 +0900)]
[moco/tf] Fix macro undef (#4210)
This will fix another wrong undef in FixPaddingTransform
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 15 Jul 2019 00:56:04 +0000 (09:56 +0900)]
[moco/tf] Introduce GraphBuilderSource interface (#4235)
This commit extracts GraphBuilderSource from GraphBuilderRegistry
which supports only lookup method.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Mon, 15 Jul 2019 00:35:23 +0000 (09:35 +0900)]
[moco_tf] helper of test to find nodes by type (#4241)
This will add find_nodes_bytype() for unit tests to find nodes by type
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Sun, 14 Jul 2019 23:33:55 +0000 (08:33 +0900)]
[moco/tf] Introduce find node by type test helper (#4226)
* [moco/tf] Introduce find node by type test helper
This will introduce find_node_bytype in test helper to find a node with a specific type
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* rename and include header
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Sun, 14 Jul 2019 23:28:34 +0000 (08:28 +0900)]
[moco/tf] Enable FeatureDecode-ReLU reorder by default (#4228)
It turns out that "FeatureDecode-ReLU" reorder transformation is
compatible with all the current implementation.
Let's enable this transformation by default.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 07:57:58 +0000 (16:57 +0900)]
[exo/tflite] Support FeatureBiasAdd (#4229)
T/F Lite exporter now accepts FeatureBiasAdd node.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 07:49:20 +0000 (16:49 +0900)]
Stop CMake immediately on GTest configure/build error (#4232)
Now, CMake stops immediately on GTest configure/build fails.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 07:49:10 +0000 (16:49 +0900)]
Stop CMake immediately on FlatBuffer configure/build error (#4231)
Now, CMake configuration stops immediately if FlatBuffer configure/build
fails.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 12 Jul 2019 05:49:54 +0000 (14:49 +0900)]
[moco/tf] Import as TFConv2D (#4223)
This will enable import as TFConv2D
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 05:32:04 +0000 (14:32 +0900)]
[exo/tflite] Export T/F Lite operation in exeuction order (#4222)
* [exo/tflite] Export T/F Lite operation in exeuction order
The current implementations generates T/F Lite operators following node
generation order which may differ from actual execution order.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Apply patch
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 05:14:26 +0000 (14:14 +0900)]
[moco/tf] Selective Reorder (#4216)
This commit allows users to control whether to apply reorder transform
per operation.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 05:14:15 +0000 (14:14 +0900)]
[exo/tflite] Remove commented function declarations (#4218)
These functions declarations are no longer used, but just commented
to reduce the amount of code diff.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 05:13:09 +0000 (14:13 +0900)]
[exo/tflite] Export Tensors after Allocation (#4199)
This commit rewrites exportOpDefinedTensors as a 2-stage function. Now,
exportOpDefinedTensors first allocates a T/F Lite tensor first, and then
export allocated T/F Lite tensors.
The first phase iterates over loco nodes, but the second phase now
iterates over allocated T/F Lite tensors.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 05:06:40 +0000 (14:06 +0900)]
[mocotest/tf] Use CMake include (#4214)
This commit revises how to add moco a T/F testcase. This change allows
users to comment out some testcases temporarily.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 04:50:01 +0000 (13:50 +0900)]
[moco/tf] Move Transform.test.cpp (#4217)
Transform.cpp is in the top-level source directory, but Transform.test.cpp
is currently inside Transforms directory.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 12 Jul 2019 04:47:07 +0000 (13:47 +0900)]
[locoex/customop] Introducing VariadicArityNode (#4189)
* [locoex/customop] Introducing VariadicArityNode
In case of custom op, we don't know how many inputs custom op has.
Because of this, previous we need this (previous FixedArityNode cannot handle this situation)
Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* unsigned to uint32_t, remove _arity member variable and duplicate assert()
* test case
* removed customop dir and namespace
* use unused var pull_02
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 04:11:43 +0000 (13:11 +0900)]
Update git ignore file (#4173)
Cleanup .gitignore file to manage files generated by official tool only
Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 12 Jul 2019 02:38:24 +0000 (11:38 +0900)]
[res] TF_SMALL_NET_0008 with Conv2D + Add (#4207)
This will introduce TF_SMALL_NET_0008 having Conv2d and Add node
This is to check FuseBinaryIntoPreceding transformation that fuses Add into Conv2D + BiasAdd
where BiasAdd will be created with 0 bias if not exist
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 00:37:14 +0000 (09:37 +0900)]
[locop] Introduce CanonicalNodeSummaryBuilder (#4205)
* [locop] Introduce CanonicalNodeSummaryBuilder
This commit introduces CanonicalNodeSummaryBuilder, which allows users
to embed CanonicalNodeSummryBuilder into their own NodeSummaryBuilder.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Update example
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 12 Jul 2019 00:14:53 +0000 (09:14 +0900)]
Stop CMake immediately on Protobuf configure/build failure (#4215)
The current implementation just returns with STATUS message.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 11 Jul 2019 09:49:03 +0000 (18:49 +0900)]
[locop] Add license notice (#4213)
There is no license notice in FormattedGraph.h and ForamttedGraph.cpp.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 06:56:59 +0000 (15:56 +0900)]
[moco/tf] Show lines in transformation log (#4190)
This will show lines in PhaseRunner for better readability in between transforms and phase
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 06:56:39 +0000 (15:56 +0900)]
[res] TF_SMALL_NET_0007 with Conv2D+Mul (#4200)
This will introduce TF_SMALL_NET_0007 having Conv2D and Mul node
This is to check FuseBinaryIntoPreceding transformation that fuses Mul into Conv2D kernel and be removed
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 06:56:11 +0000 (15:56 +0900)]
[moco/tf] TFConv2D to Conv2D Canonicalizer (#4202)
This will implement TFConv2D to Conv2D Canonicalizer
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 06:55:44 +0000 (15:55 +0900)]
[moco/tf] Fix macro undef (#4201)
This will fix macro undef to right name
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 11 Jul 2019 06:17:50 +0000 (15:17 +0900)]
[locop] Extract canonical_node_desc from node_desc (#4198)
This commit extracts canonical_node_desc helper from node_desc in order
to make it easy to reuse.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 11 Jul 2019 04:59:37 +0000 (13:59 +0900)]
[tf2tflite] Enable Placeholder_000 and Multiple_IO_002 (#4196)
Recent changes on moco-tf and exo-tflite resolve Placeholder_000 and
Multiple_IO_002 test failure.
- moco-tf now eliminates Forward by default, which resolves
Placeholder_000 failure.
- exo-tflite now honor the input/output order, which resolves
Multiple_IO_002 failure.
This commit updates tf2tflite test framework to check Multiple_IO_002
and Placeholder_000 by default to prevent regression.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 03:47:45 +0000 (12:47 +0900)]
[moco/tf] Use PadData and StrideData for TFConv2d (#4163)
* [moco/tf] Use PadData and StrideData for TFConv2d
This will fix TFConv2D related FixPadding and FixShare transformations to use PadData and ShapeData to simplify Canonicalization
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* check data_layout
* fix log message
* fix log message
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 01:37:44 +0000 (10:37 +0900)]
[moco/tf] Add fix for shape and padding (#4188)
This will add fix for shape and padding after TF transformations that can add new nodes with empty shape and padding
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 11 Jul 2019 01:35:28 +0000 (10:35 +0900)]
[moco/tf] Remove unused Canonicalization files (#4191)
This will remove unused Canonicalization files that are replaced by the files in Canonicalization folder
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 11 Jul 2019 01:21:33 +0000 (10:21 +0900)]
[exo/tflite] Hide ShapeContext (#4187)
This commit makes ShapeContext visible only to ShapeInference module,
and revises all the other modules to use "ShapeInferece::get".
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 11 Jul 2019 00:33:39 +0000 (09:33 +0900)]
Add Multiple_IO_002 TensorFlow test (#4181)
This commit adds Multiple_IO_002 test. This testcase allows us to
check what happens if there is a mismatch between node declataion order
and input order.
Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>