saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-06-04 23:59:56 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
645 commits 2856 branches 57 tags 1.3 GiB
Commit graph

645 commits

This branch
This branch
All branches
Author SHA1 Message Date
Pranav Sharma
049ba2d747
Exclude tests that fail when contrib ops are disabled. (#835) 2019-04-15 15:57:48 -07:00
Pranav Sharma
4b4a359943
Exclude unreferenced global data and op doc strings in the opschema object. The first causes a decrease in the binary size by at least 85k. The latter reduces resident memory size. (#823) 
* Exclude unreferenced global data and op doc strings in the opschema object. The first causes a decrease in the binary size by at least 85k. The latter reduces resident memory size.

* Update onnx to incorporate my PR that fixes SetDoc compiler warnings
 2019-04-15 15:57:19 -07:00
Ashwini Khade
e999af61b2
bug fix for shape inference (#834) 2019-04-15 15:51:12 -07:00
Raymond Yang
fabdbdc130
Update test retrieval following #828 (#836) 
* Enable nightly build

* Update fetch file names

* Fix

* Update setup.py

* Update run_dockerbuild.sh

* Resolve comments

* Update test data
 2019-04-15 14:51:20 -07:00
Dmitri Smirnov
6194a92249
Fix empty input handling in Tokenizer. (#826) 2019-04-15 14:46:17 -07:00
Changming Sun
2c0b8e965e Disable test data local cache in Linux CI pipelines 2019-04-12 22:23:16 -07:00
Changming Sun
e493ba2219 Fix memory leaks in perf test runner 2019-04-12 00:51:33 -07:00
Raymond Yang
1936d141a7
Create nightly build for python packages (#817) 
* Enable nightly build

* Update fetch file names

* Fix

* Update setup.py

* Update run_dockerbuild.sh

* Resolve comments
 2019-04-11 22:06:18 -07:00
Tracy Sharpe
c55e2de593
Status class optimizations (#824) 
optimize onnxruntime::common::Status to reduce code size
 2019-04-11 21:57:01 -07:00
Hariharan Seshadri
b6936e71cb
Avoid postfix iterator increment in a loop in Slice op and some minor formatting fixes (#820) 
* Initial commit

* Fix comment

* More nit fixes
 2019-04-11 17:44:32 -07:00
Hariharan Seshadri
ccf3566c35 Register kernel for Dropout (opset 10) for opset compliance (#813) 2019-04-11 13:55:43 -07:00
Pranav Sharma
6577c3dddf
Extract debug symbols in a separate file and strip the binary. (#811) 
* Ensure Linux binaries are built with debug info. Extract debug info out of the main binaries. Strip the main binaries.

* add binutils

* add uname

* add binutils

* remove linux portion
 2019-04-11 12:02:50 -07:00
Ryan Hill
1ff29bfb3d
Fix x86 calling convention break (#814) 2019-04-11 10:41:07 -07:00
Hector Li
0741baf867
Update NMS to support max_output_boxes_per_class = 0. NMS will do nothing for this case. (#816) 2019-04-11 10:09:33 -07:00
Hariharan Seshadri
56749a84ee
Implement opset v10 changes for Slice operator (#772) 2019-04-10 22:06:05 -07:00
jignparm
53038b33ed BuildFusedKernelDef uses N^2 algorithm verifying input constraints; session load time is huge for fused nodes (#804) 
This optimization is required for WinML to prevent unit test time out.
 2019-04-11 09:52:10 +08:00
jignparm
d17ae5c093
MKLML pipeline - update C# and CMake to handle dll dependencies (#810) 
* Refactor NuGet to allow arbitrary namespaces

* Move csharp build to end of cmake

* Minor edit to ensure dll generation in sequence
 2019-04-10 18:16:02 -07:00
Jesse Benson
24d80b4bda Add support for BrainSlice execution provider in Python, if onnxruntime is built with it. 2019-04-10 17:37:21 -07:00
Ashwini Khade
10b113f144
update onnx to bring in quantized ops (#808) 
* update onnx + move quantized ops kernels and test to onnx + remove exp ops

* update onnx

* Revert "update onnx"

This reverts commit 533abfc297e75473a74505fb89921ffc05c46a1c.

* add generated csharp test file
 2019-04-10 17:20:35 -07:00
Changming Sun
4bc3d6027d Build perf test runner only if onnxruntime_BUILD_SHARED_LIB is ON 2019-04-10 13:16:56 -07:00
Raymond Yang
3dcf82a1f9 Disable some flaky tests with CUDA9 (#805) 
* Disable failing cuda tests
 2019-04-10 01:02:32 -07:00
jignparm
4e3391ef60
Refactor NuGet to allow arbitrary PackageId names (e.g. Microsoft.ML.OnnxRuntime.MKLML) (#797) 
* Refactor NuGet to allow arbitrary namespaces

* Move csharp build to end of cmake
 2019-04-09 22:48:00 -07:00
Ashwini Khade
e7090d7202
move all removed exp ops to contrib ops (#786) 
* move all removed exp ops to contrib ops

* fix cuda build failure

* bug fix

* move some tests to contrib ops + cosmetic changes

* Revert "move some tests to contrib ops + cosmetic changes"

This reverts commit 4cda9297e257a6f6b902724e8113bf5d5a62df29.
 2019-04-09 22:26:48 -07:00
Changming Sun
0d4055def4 Integrate tensorflow into onnxruntime_perf_test tool 2019-04-09 15:55:08 -07:00
jignparm
9467c5f967
Update version to 0.3.1 (patch release) (#798) 
* bump up version number (#752)

* bump up version number

* Minor change to kick off build

* update version to 0.3.1
 2019-04-09 14:48:56 -07:00
Xavier Dupré
ccd7e801a0 Fix #612, TfidfVectorizer handles empty matrices as an input (#702) 
* Fix #612, TfidfVectorizer handles empty matrices as an input

* Add more unit tests, better consistency of error messages

* Update tfidfvectorizer.cc

* better comment

* fix comments

* add unit test failure for an empty input {0, 1}
 2019-04-09 10:55:24 -07:00
Yufeng Li
39951f35f4
Use template windows-build-tools-setup-steps.yml in win pipelines (#794) 
1.  Update nuget restore to 4.3 for capi pipeline
2. Use template windows-build-tools-setup-steps.yml in win piplines.
 2019-04-08 21:35:33 -07:00
jywu-msft
d91555f99e
fix for tensorrt_basic_test not being run. (#792) 2019-04-08 13:18:36 -07:00
Hariharan Seshadri
5cf72030b2
Rename misleading test names in ConvTranspose op tests (#788) 2019-04-06 17:01:26 -07:00
jywu-msft
571291c323 build.sh: don't require user to set --use_full_protobuf with --use_tensorrt option. we can set it implicitly. (#780) 
* use_full_protobuf if tensorrt build option is enabled.

* update BUILD.md sections on MKLDNN and TensorRT/full_protobuf option
 2019-04-06 10:11:57 -07:00
Yufeng Li
cea2a40bf1
Clean up ExecutionProvider in CSharp (#783) 2019-04-05 22:29:54 -07:00
Ryan Hill
fda1d0dce9
Ryanunderhill/ocr custom op (#744) 
* Adding a custom op interface to the C API to remove shared library dependency.
* Remove old custom op test
* Rework how custom ops handle inputs/outputs to enable custom op output shape calculation in the compute method
* Add a nicer C++ API for custom ops and switch the tests to use it.
 2019-04-05 18:53:20 -07:00
Tao Qin
58ef1306d4
Copy inputs and outputs directly in InferenceSession::SaveModelMetadata (#777) 
* Copy required inputs and outputs directly in InferenceSession::SaveModelMetadata

* trivial

* trivial
 2019-04-05 15:16:55 -07:00
ybrnathan
3eddb2d61e
Add optimization level as cmd line arguments (#776) 
* Add optimization level as cmd line arguments

* fix the help info and add option.
 2019-04-05 14:44:28 -07:00
utsabsingharoy
36ed91ee9f CustomRegistry should use composition instead of inheritence 
CustomRegistry should use composition instead of inheritence
 2019-04-05 14:14:10 -07:00
Changming Sun
867e961ee8 Remove mkldnn_sgemm from math_util.cc 
If it is needed, it can be used explicitly in mkldnn provider.
 2019-04-05 14:13:10 -07:00
Hariharan Seshadri
ffd9071168
expose graph node name returning non-zero status code (#714) 
* Initial commit
 2019-04-05 12:50:58 -07:00
Peng Wang (AI FWK)
f4021cf30a fix a minor inaccurate error message 2019-04-05 12:07:28 -07:00
Yufeng Li
ef9a4d98cb
Expose parallel execution option in C# API (#767) 
* Expose parallel execution option

* delete unnesary file

* add doc

* update nuget retore to 4.3.0

* resolve comments

* remove unnessary file

* make git ignore csharp/Directory.Build.props

* fix yaml config for nuget 4.3
 2019-04-05 12:05:56 -07:00
Changming Sun
43521c0de7 update 2019-04-04 21:22:27 -07:00
Scott McKay
65c50bb25b
Create NodeArg for all initializers if IR version is > 3. (#742) 
Previously all initializers had to have matching graph inputs and the NodeArg was guaranteed to be created via graph input processing.
 2019-04-05 14:09:27 +10:00
Mika Fischer
2674d9bd8a Fix profiling with C API (#710) 
Currently, when using OrtEnableProfiling to enable profiling using the C API,
the profile output file is created but is always empty.

The reason is that InferenceSession::EndProfiling() needs to be called to
write the profiling data to the output file.

However there's currently no way to call this function via the C API.

This adds a call to EndProfiling() to the descructor of the session if
profiling is enabled in the session options.
 2019-04-04 16:23:18 -07:00
Raymond Yang
c35d9797d7
Add unique identifier in function subgraph (#771) 
* Add unique identifier in function subgraph

* enable test
 2019-04-04 13:31:20 -07:00
Changming Sun
289663fd85 reformat onnxruntime_c_api.cc 2019-04-04 13:03:13 -07:00
Changming Sun
290112d614
Update onnx (#761) 
* update onnx
 2019-04-04 10:58:45 -07:00
Changming Sun
261a9078a5 Remove redundant code in ml::write_scores 2019-04-04 01:21:48 -07:00
Konstantinos Karanasos
512cfdd9fe
Generalize node removal (#743) 
Generalize node removal method in graph_utils. This is a higher-level method that keeps the graph consistent so that no Resolve is needed after the removal of a node. 
The new method supports the removal of nodes with a single input (be it an incoming node or an initializer) and a single output (but allowing multiple output edges of that output). It also takes into account the case that one of the output edges is fed to a subgraph.
Also updated the rewrite rules to use this new, less restrictive method, and improved the rules' conditions. Introduced a GraphEdge struct to simplify various methods in graph_utils.
 2019-04-03 22:34:20 -07:00
Changming Sun
7af35ac1e6 Fix two warnings in graph lib 2019-04-03 19:41:11 -07:00
Changming Sun
a196085471 Add error messages to FunctionImpl::FunctionImpl 2019-04-03 19:40:49 -07:00
Shawn Callegari
adc00ab5fc ARM32v7 Dockerfile and build instructions update. (#737) 2019-04-03 14:45:24 -07:00