saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-03 03:58:54 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
3515 commits 2856 branches 57 tags 1.3 GiB
Commit graph

3515 commits

This branch
This branch
All branches
Author SHA1 Message Date
Pranav Sharma
c2c78399ee
Include config keys header file in the release packages for Linux and Mac. (#5388) 2020-10-08 15:00:29 -07:00
Changming Sun
09aef240d6
Skip running onnx tests in python mac os pipeline (#5416) 2020-10-08 11:49:28 -07:00
Tiago Koji Castro Shibata
83ead3e2eb
Fix com ptr refcount (#5404) 2020-10-08 10:18:38 -07:00
Yufeng Li
b04cf2d229
Update ORT to 1.5.1 in Bert Quantization Notebook (#5396) 
* Update ORT to 1.5.1 in Bert Quantization Notebook
 2020-10-08 09:55:01 -07:00
manashgoswami
132ab2230d
Updated with image for creating the onnxruntime pkg (#5400) 
* Create Mobile.png

* Update ONNX_Runtime_for_Mobile_Platforms.md

* Update ONNX_Runtime_for_Mobile_Platforms.md
 2020-10-08 08:54:27 -07:00
Scott McKay
9684e1b5a8
Add doco for pre-requisites to be able to cross compile for Android on Windows with Java bindings enabled. (#5395) 2020-10-08 12:31:46 +10:00
Tianlei Wu
8133223871
clear cudaDelayLoadedLibs since delayload is disabled (#5386) 2020-10-07 11:33:12 -07:00
Tianlei Wu
8ee2b08325
Allow benchmark different threads (#5390) 2020-10-07 11:13:01 -07:00
Tianlei Wu
094384781e
Add --use_external_data_format in convert_to_onnx.py (#5393) 2020-10-07 09:42:02 -07:00
Guoyu Wang
5947445457
Add flatbuffers verifier for ORT format buffer (#5378) 
* Add flatbuffers verifier before accessing data in ort format models

* Address review comments
 2020-10-07 09:23:17 -07:00
Guoyu Wang
deb708d3b1
Move flatbuffers to 1.12 release (#5392) 2020-10-07 09:23:03 -07:00
Hariharan Seshadri
6f54113a1b
Support OrtValue binding in Python to enable interesting IOBinding scenarios in Python (#5248) 2020-10-06 21:14:41 -07:00
Tracy Sharpe
0122e890d9
MLAS: implement u8x8 GEMM for ARM64 (#5380) 
Add an implementation for u8u8/u8s8 GEMM for use on ARM64 (Windows/Linux).
 2020-10-06 19:22:23 -07:00
Guoyu Wang
b4934b0016
Mitigate pybind11 build break using Xcode 12 on macOS (#5381) 
* turn dev_mode off if we are using macos to build python with xcode 12

* Address CR comments

* Add ways to check compiler version
 2020-10-06 19:03:33 -07:00
Kaarthik Sivashanmugam
10f1902d90
Update code snippet in README.md 2020-10-06 17:41:56 -07:00
liqunfu
773992c7d4
Liqun/bert pretrain tb (#5377) 
* add tensor board, remove torch.distributed.lanuch because ort nccl depends on MPI. Use MPI to launch parallel training.

Co-authored-by: liqun <liqun@OrtTrainingDev4.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2020-10-06 16:28:31 -07:00
manashgoswami
b5caa7cb12
Updated docs: Execution Provider overview (#5328) 
* Update ReleaseManagement.md

* Create ONNX_Runtime_Execution_Providers.md

* Create ONNX_Runtime_EP3.png

* Create ONNX_Runtime_EP2.png

* Create ONNX_Runtime_EP1.png

* Delete ONNX_Runtime_Execution_Providers.md

* Create README.md

* Update README.md

* commit

* Updated in error.
Revert "Update ReleaseManagement.md"

This reverts commit 8530bd5fd46aebce3a6d6055d8952ae4f6458c4e.

* Create ONNX_Runtime_Execution_Providers.md

* Create ONNX_Runtime_EP3.png

* Create ONNX_Runtime_EP2.png

* Create ONNX_Runtime_EP1.png

* Delete ONNX_Runtime_Execution_Providers.md

* Create README.md

* Update README.md

* commit

* Updated in error.
Revert "Update ReleaseManagement.md"

This reverts commit 8530bd5fd46aebce3a6d6055d8952ae4f6458c4e.

* Update ReleaseManagement.md

* Update .gitignore

* Update README.md

* Update README.md
 2020-10-06 15:01:25 -07:00
Du Li
323c4dfe02
Adding an option for cudnn conv algorithms. (#5159) 
* adding cudnn conv algorithm selection options.

* adding cudnn conv algorithm selection options.

* export the api

* adding the perf test option.

* accomodating pr comments.

* Move OrtSessionOptionsAppendExecutionProvider_CUDA to onnxruntime_c_api.h

* Accomodating PR comments.
 2020-10-05 16:53:52 -07:00
Shucai Xiao
a0b8218f9a
Amdmigraphx update to rocm3.7 (#5362) 
* backup dockerfile for upgrading to rocm3.7

* fix build errors related to rocm3.7

* backup dockerfile for migraphx

* remove unnecessary component from dockerfile

* fix review comments

Co-authored-by: Shucai Xiao <scxiao@prj47-rack-99.local.lan>
 2020-10-05 15:34:24 -07:00
Yufeng Li
24f99b3be8
Support OuterStride for QGemm when MLAS_SUPPORTS_GEMM_U8X8 undefined (#5374) 
Quantized GEMM on ARM doesn't support the case that leading dimension is not equal to column size. The PR adds support of this case.
 2020-10-05 13:06:12 -07:00
Ashwini Khade
668ab04917
rename all TransposeMatMul nodes to FusedMatMul (#5373) 2020-10-05 12:41:05 -07:00
Wei-Sheng Chin
4e3a420aa7
Use single thread when pipeline is not enabled in TrainingRunner (#4265) 
* Use single thread when pipeline is not enabled in TrainingRunner

* Remove macro indents

* Format file and remove state variable
 2020-10-05 10:42:09 -07:00
Vlad Burlik
c20fcf26eb
Onnx GPU runtime fails to fallback to CPU when GPU is not available/busy (#5304) 
* ONNX GPU runtime fails to fallback to CPU when GPU is not available OR busy
https://github.com/microsoft/onnxruntime/issues/5299

* comments

* Init _fallback_providers before C.InferenceSession

* As per review: Fallback providers order supersedes user's providers order, IF they are included into providers list.

* Code convention fix

* pep8
 2020-10-02 22:45:14 -07:00
Wenbing Li
4721729fdc
Enable iOS CI pipeline (#5360) 
* add the ios ci build.

* no dependency on mac ci pipeline.

* fix the command line.

* keep sync

* automatically retrieve sdpath

* fix the case errors and warnings

* fix the vlog switch issue.

* add parallel flag for build.

* update the display name of the pipeline.
 2020-10-02 20:14:45 -07:00
Guoyu Wang
9df0790856
Update linux minimal CI to report Android mininal baseline binary size (#5361) 
* Update linux minimal CI to report Android mininal baseline binary size

* Fix some issues in the script
 2020-10-02 17:35:23 -07:00
Chun-Wei Chen
5bd7241839
Raise output mismatch error in ort_test_dir_utils.py (#5364) 2020-10-02 16:44:59 -07:00
Tianlei Wu
f5e4c0ea04
Fix benchmark_gpt2 model verification (#5343) 2020-10-02 13:53:02 -07:00
Guoyu Wang
6e4949e235
javadoc warning fix (#5332) 
Co-authored-by: gwang0000 <62914304+gwang0000@users.noreply.github.com>
 2020-10-02 11:52:07 -07:00
Hariharan Seshadri
06cd81d791
Support trilinear sampling in Resize CPU and CUDA kernels (#5300) 2020-10-02 11:02:43 -07:00
Sherlock
e71668f92c
Expose recompute configs to the frontend (#5318) 
* Expose recompute configs to the frontend

* Add frontend test

* Ensure recompute graph transformer is only applied once

Co-authored-by: Sherlock Huang <bahuang@OrtTrainingDev3.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2020-10-02 09:49:47 -07:00
Tianlei Wu
e33de20861
Update gpt2 notebook for int8 quantization (#5346) 
* Update gpt2 notebook for ORT 1.5
* add sections for int8 quantization including QAT note
 2020-10-02 09:41:52 -07:00
Ashwini Khade
ce49cfa67c
add support for configurable build dir when building nuget packages (#5352) 
* add support for configurable build dir when building nuget packages

* rename vars
 2020-10-02 09:31:35 -07:00
Changming Sun
f265834c2c
Exclude GPT2_LM_HEAD from OpenVino's model test list (#5356) 
GPT2_LM_HEAD is a new ONNX model zoo model that OpenVino doesn't support.

Error message:1: [ONNXRuntimeError] : 6 : RUNTIME_EXCEPTION : Non-zero status code returned while running OpenVINO-EP-subgraph_1162 node. Name:'OpenVINOExecutionProvider_OpenVINO-EP-subgraph_1162_1' Status Message: _Map_base::at
 2020-10-01 21:49:45 -07:00
Sunghoon
1612934f72
Allow protobuf format of input data for performance test (#5323) 
* Allow protobuf format of input data like onnxruntime_perf_tool

* Add OnnxML.cs to fix build failure
 2020-10-01 21:40:29 -07:00
Yufeng Li
e8b9aa1f29 fix quantization of EmbeddingLayerNorm (#5321) 2020-10-01 20:08:43 -07:00
KeDengMS
7495dc167a
Symbolic shape inference: fix a bug in auto_merge when broadcasting (#5349) 
The bug happens when merging following shapes:

input0: [1, 1, 'Min(1024, input1_dynamic_axes_3)', 'Min(1024, input1_dynamic_axes_3)']
input1: ['input1_dynamic_axes_1*input1_dynamic_axes_2', 12, 'input1_dynamic_axes_3', 'input1_dynamic_axes_3']
input2: []

The fix is to avoid broadcasting merge on input2
 2020-10-01 15:24:00 -07:00
Ye Wang
caed6c264c
Add tf2pytorch wrapper in transformers tool (#5316) 
* init checkin

* format

* refactor

* review comments
 2020-10-01 13:58:58 -07:00
edgchen1
d62873a331
Docker image release build updates (#5326) 
- Update docker image release build to use build commit.
- Use valid default in component governance detection step.
- Use smaller docker build context.
 2020-10-01 12:25:31 -07:00
liqunfu
fe50213491
Liqun/bert pretrain2 (#5327) 
* bert single node multi GPU pretrain w/o checkpoint

Co-authored-by: liqun <liqun@OrtTrainingDev4.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2020-10-01 11:01:26 -07:00
Brian Martin
1cad3e322e
typo in contributing.md (#5340) 
there's a missing space between two words.
 2020-10-01 10:23:08 -07:00
Guoyu Wang
2098d621a6
Make some string optional for save to/load from flatbuffers (#5331) 
* Update how to save and load string using flatbuffers and ort_format_only_test

* Add some comments

* Address PR comments
 2020-10-01 09:24:37 -07:00
Hariharan Seshadri
383b1e207c
Fix bug in the Resize operator kernels (#5303) 2020-09-30 15:33:33 -07:00
Ashwini Khade
3f00b8db8f
move all experimental ops to version 1 of ms domain (#5287) 
* move all experimental ops to version 1 of ms domain

* deprecate TransposeMatMul in favor of FusedMatMul

* update documentation
 2020-09-30 14:50:18 -07:00
edgchen1
2c32309e2c
Update dockerfiles/README.md onnxruntime-training image tags. (#5333) 2020-09-30 14:35:38 -07:00
Sherlock
37445d1198
Update Bert Perf Script (#5339) 
Co-authored-by: Sherlock Huang <bahuang@OrtTrainingDev3.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2020-09-30 14:30:20 -07:00
Changming Sun
8d4740b39c
Add some log for the GetFileLength function (#5330) 2020-09-30 10:39:42 -07:00
Faith Xu
cb57c100e6
Doc updates for 1.5 (#5302) 
* Fix Windows AI version

* Update text to extend telemetry coverage 

Includes all official binaries

* Update text about EP pluggability

* Update CUDA/cuDNN versions

* Add link to reduce operator kernel page

* Update roadmap

* Add preview for migraphx

* Move Rockchip under IoT/Edge

* Update text to include ORT for Mobile doc link
 2020-09-30 09:53:33 -07:00
Tim Harris
69dbaaa015
Add additional test cases to check for leaks in thread pool creation / destruction (#5311) 
Add additional test cases such as ThreadPoolTest.TestPoolCreation_10Iter to create and destroy thread pools to watch for any memory leaks. Running under Valgrind, these tests should show all of the data allocated being deallocated again.

Two recent issues #5176 and #5292 indicated memory leaks. The test cases help identify whether or not any of the data structures used in the thread pool are being leaked. Currently, on WSL, the only data not being de-allocated in these tests are a small number of nsync waiter objects. This behavior is as expected (the waiter objects should be held on a free list in the nsync library).
 2020-09-30 11:26:02 +01:00
Ye Wang
1a12f510fc
Support T5 benchmarking in transformers tool (#5133) 
* init checkin

* review comments

* modify according to transformers release
 2020-09-29 22:58:28 -07:00
Sherlock
9ec1ed42a8
Enable BiasDropoutFusion for CUDA EP only (#5324) 
Co-authored-by: Sherlock Huang <bahuang@OrtTrainingDev3.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>
 2020-09-29 14:00:15 -07:00