saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-07 04:39:07 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
7863 commits 2856 branches 57 tags 1.3 GiB
Commit graph

7863 commits

This branch
This branch
All branches
Author SHA1 Message Date
Wei-Sheng Chin
de9da123cf
Enable static memory planning for pipeline. (#4204) 
* Enable static memory planning for pipeline.
1. We fix a bug when resolving symbolic shape for scalars.
2. We pass the original inputs to all pipeline stages so that
   the symbolic shapes can be resolved.

* Further Improvements
1. Address comments.
2. Further reduce activation size by ~50% when pipeline is on.
   This is done by removing all but one gradient tensor from the last
   RecordEvent in the backward pass.

* Address a comment

* Fix Windows build
 2020-06-12 21:43:50 -07:00
Hariharan Seshadri
b377266eb3
Fix Mac build linker warnings (#4155) 2020-06-12 21:10:12 -07:00
Hariharan Seshadri
91a41298cc
Fix ORT build when onnxruntime_PYBIND_EXPORT_OPSCHEMA is enabled (#3954) 2020-06-12 19:32:57 -07:00
Tracy Sharpe
155e22d1ab
MLAS: fuse float output into quantized GEMM (#4215) 
Add more variants of MlasGemm that do a u8x8 GEMM with the output type as float. This fuses the common sequence of MatMulInteger + Cast + Mul(OutputScale) + optional Add(BiasVector).
 2020-06-12 17:50:40 -07:00
Tiago Koji Castro Shibata
2e3607c7cd
Remove hardcoded desktop lib (#4193) 2020-06-12 16:51:54 -07:00
ISS Build Account
01edd61372 Merge remote-tracking branch 'upstream/master' into DmlDev 2020-06-12 23:12:57 +00:00
Edward Chen
f74861841e Fix dangling pointer to local string variable in onnxruntime_pybind_state.cc. 2020-06-12 14:28:39 -07:00
Edward Chen
6b4f652017 Clean up status checks in gradient_graph_builder_test.cc. 2020-06-12 14:28:39 -07:00
Edward Chen
7096e6f5ef Reduce severity of GraphAugmenter logging statement. 2020-06-12 14:28:39 -07:00
Changming Sun
6f4320fb85
Fix the python package name issue (#4207) 
Fix the package package name issue. In my last change(#4197) about enabling code sign. I forgot to pass the additional flags to setup.py,
 2020-06-12 08:32:59 -07:00
Yufeng Li
87d68d8531
matmul integer fusion (#4195) 
* Introduce DynamicQuantizeMatMul
It fuses DynamicQuantizeLinear, MatMul and following cast, multiplier. It gets float in and float out for quantized matmul. We have a MLAS kernel in implementation for this op.
 2020-06-11 21:42:09 -07:00
Tianlei Wu
2605faef88
Add past state support in Attention Op for GPT-2 (#4107) 
Update Attention op to allow past state input and output.
Add fusion script and tests
 2020-06-11 14:19:55 -07:00
pengwa
e6ccb1ac28
GatherNDGrad for CPU (#4123) 
* GatherNDGrad on CPU

* Remove __CUDA_ARCH__ check in .cc files
 2020-06-12 02:43:49 +08:00
ISS Build Account
56a2dfd7d5 Merge remote-tracking branch 'upstream/master' into DmlDev 2020-06-11 17:24:00 +00:00
Xueyun Zhu
65a682354b
enable pipeline to run with mixed precision (#4113) 
* enable pipeline to run with mixed precision

* address feedback

* address feedback

* test log

* pipe infomation if test fails

* ci failure
 2020-06-10 22:16:24 -07:00
Changming Sun
8f8d899bf2 Enable code sign in c api pipeline and python pipeline 2020-06-10 19:31:22 -07:00
Yulong Wang
73bc6be5d1
build: split nodejs binding build and test to avoid timeout issue (#4188) 
* split nodejs binding build and test

* enable nodejs tests
 2020-06-10 19:16:32 -07:00
Matthew Hill
117b2e7743
Fix GPU memory leak on TensorRT (#4172) 2020-06-10 16:56:51 -07:00
Tiago Koji Castro Shibata
e323411e31 Merged PR 4790197: Remove hardcoded desktop lib 
Remove hardcoded desktop lib

Related work items: #26800043
 2020-06-10 23:16:56 +00:00
ISS Build Account
eae841464c Merge remote-tracking branch 'upstream/master' into DmlDev 2020-06-10 18:19:40 +00:00
Dmitri Smirnov
af0750ba1b
Java GPu artifact naming (#4179) 
Modify gradle build so artifactID has _gpu for GPU builds.
  Pass USE_CUDA flag on CUDA build
  Adjust publishing pipelines to extract POM from a correct path.

Co-Authored-By: @Craigacp
 2020-06-10 11:15:48 -07:00
George Wu
e8ed14bcb3 disable MEMLEAK CHECKER for openvino 2020-06-10 11:12:17 -07:00
stevenlix
c296884fc3
bump up ORT version to 1.3.1 (#4181) 2020-06-10 08:44:03 -07:00
Changming Sun
c0bdbc0b39
Enable telemetry for the C API and python pipeline (#4174) 2020-06-10 00:07:46 -07:00
Tracy Sharpe
35d9f396c4
MLAS: refactor quantized GEMM loops (#4182) 2020-06-09 23:28:55 -07:00
George Wu
9d65ce53bc
move back to toolset 14.16 to possibly work around nvcc bug (#4180) 2020-06-09 19:36:30 -07:00
Changming Sun
a7366d82af
Disable nuphar large model test (#4173) 
Disable nuphar large model test, because it takes too long(40+ minutes), while the default cpu provider takes about 5 minutes. After this change, we still keep a lot of other nuphar model tests, I think that should be enough.
 2020-06-09 17:45:17 -07:00
ISS Build Account
a47158e3ee Merge remote-tracking branch 'upstream/master' into DmlDev 2020-06-09 22:42:20 +00:00
Ashwini Khade
9eba9fba7c
Fix for BiasGelu fusion optimizer (#4160) 
* Fix for BiasGelu fusion optimizer

* changes per review comments
 2020-06-09 14:33:34 -07:00
Yulong Wang
2b3ce1b090
add script to support update nodejs binding version (#4164) 2020-06-09 13:12:55 -07:00
ISS Build Account
559d2c0c9c Merge remote-tracking branch 'upstream/master' into DmlDev 2020-06-09 17:59:03 +00:00
Tiago Koji Castro Shibata
a9862b00f3 Merged PR 4779170: Merge ORT master 2020-06-09 17:51:10 +00:00
Sheil Kumar
4377ff4a1a
Enable .NET Core 2.0 and .NET Framework 4.6.1 in Microsoft.AI.MachineLearning NuGet package (#4125) 
* add project to download cswinrt and build winrt c# interop dll

* Add to nuget package

* reverse if check

* run generation before core compile

* add generated files to compile

* update .net package to binplace native libs

* add props to .netstandard2.0 folder

* auto binplace ml native binaries

* force 'Any CPU' platform build

* Fix anycpu and platform targets

* fix flake errors

* fix variable order

* fix flake pep8 errors, semicolon

Co-authored-by: Sheil Kumar <sheilk@microsoft.com>
 2020-06-09 09:08:19 -07:00
Scott McKay
28d12dc4f0
Try to avoid std::move in return whilst keeping CentOS build happy. (#4163) 2020-06-09 21:41:49 +10:00
oak-tree
541eafb41a
Fixed the link to model test documenation (#4011) 2020-06-08 17:27:55 -07:00
Changming Sun
2ab3a19728
Enlarge the read buffer size in C#/Java test code (#4150) 
1. Enlarge the read buffer size further, so that our code can run even faster. TODO: need apply the similar changes to python some other language bindings.
2. Add coreml_VGG16_ImageNet to the test exclusion set of x86_32. It is not a new model but previously we didn't run the test against x86_32.
 2020-06-08 16:13:11 -07:00
Tiago Koji Castro Shibata
045ce7d13b Merge remote-tracking branch 'upstream/master' into user/ticastro/update-DmlDev 2020-06-08 15:33:33 -07:00
Tiago Koji Castro Shibata
8eb6a539bd
Hardcode WinML tests umbrella lib (#4161) 2020-06-08 15:24:08 -07:00
suffiank
7f5339505e
Discover trainable parameters using reverse DFS from loss node (#4116) 
Discover trainable parameters using reverse DFS from loss node, omitting recursion along untrainable inputs.

Co-authored-by: suffian khan <sukha@OrtTrainingDev4.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
Co-authored-by: suffian khan <sukha@microsoft.com>
 2020-06-08 14:16:10 -07:00
Yulong Wang
842be1535d
[Node.js binding] add linux and mac package (#4157) 
* try mac pipeline

* fix path separator

* copy prebuilds folder

* split esrp yaml for win/mac

* disable mac signing temporarily

* add linux

* fix indent

* add nodetool in linux

* add nodetool in win-ci-2019

* replace linux build by custom docker scripts

* use manylinux as node 12.16 not working on centos6

* try ubuntu

* loosen timeout for test case - multiple runs calls
 2020-06-08 14:12:05 -07:00
Sergii Dymchenko
653417ae4b
Fix scaler->scalar typo. (#4142) 2020-06-08 13:02:12 -07:00
Tiago Koji Castro Shibata
6bbd18efd0
Hardcode WinML umbrella lib to windowsapp.lib (#4133) 2020-06-08 11:04:44 -07:00
Wenbing Li
ee35320974
The fixings for python scripts in ONNXRuntime (#4135) 
* The fixings for python scripts in ONNXRuntime

* update according the comments
 2020-06-08 10:27:32 -07:00
Faith Xu
3390431d80
Update MCR image table (#4137) 2020-06-08 10:13:13 -07:00
Changming Sun
5a5f44eed7
Add softmax to the mnist example (#4149) 2020-06-08 09:33:50 -07:00
Dmitri Smirnov
4e1dac67cd
Address memory leak and improve memory handling (#4124) 
Fix memory leak when a Python list passed as a feed.
  Create a custom allocator that can take ownership of python
  arrays that are created inside pybind.
  Allow direct memory use if continuous array is a copy because
  we now can take ownership of it by the allocator.
 2020-06-08 09:29:46 -07:00
Cecilia Liu
b8db8076cb
Fix MKLML Tests Run (#4144) 
Add a path to LD_LIBRARY_PATH to fix library not found error when running mklml test cases.
 2020-06-06 20:28:53 -07:00
Tianlei Wu
7c8e1580a1
Add check of graph output in Bert Fusions (#4126) 
* Refine node output edge checking in bert related fusions
 2020-06-06 00:06:07 -07:00
liqunfu
ffed43e9b8
handle loss and name marching wrappers (#4066) 
* handle loss and name marching wrappers

Co-authored-by: liqun <liqun@OrtTrainingDev4.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2020-06-05 23:34:26 -07:00
Yulong Wang
2aab20b4ea
[Node.js binding] upgrade node-addon-api to 3.0 (#4148) 2020-06-05 21:24:34 -07:00