microsoft/onnxruntime

Pull Requests Commits

Fix bug where linear_output_ is not cleared when linear_before_reset is true and no bias input is provided. Requires a batch size of 3 or more to trigger if initial_h is not provided. (#3893)

skottmckay committed 5 years ago

Verified 99415f09

Fix ACL EP convolution-activation fusion optimization (#3896)

Andrews548 committed 5 years ago

Verified 475ea382

Extend workaround with input name matching in DML fused graph kernel (#3918)

jeffbloo committed 5 years ago

Verified f170f31e

Ensure pt model is set to cpu in ort_trainer (#3867)

BowenBao committed 5 years ago

Verified 0f82b42f

Add Benchmark Script for Bert Models (#3829)

tianleiwu committed 5 years ago

Verified 29496170

Fix bug in handling of an initializer that provides a graph output. (#3912)

skottmckay committed 5 years ago

Verified 2fed37c8

use LOAD_WITH_ALTERED_SEARCH_PATH for LoadLibraryExA (#3908)

jywu-msft committed 5 years ago

Verified 6f729b10

Remove ORTModel from frontend API (#3825)

Thiago Crepaldi committed 5 years ago

Verified 70abb120

MLAS: tune softmax kernels for partial vectors (#3906)

tracysh committed 5 years ago

Verified b12d35b6

Add numerical stability to SoftmaxGrad test inputs. (#3857)

codemzs committed 5 years ago

Verified c46a9e8d

Merge DML Execution Provider updates (#3885)

jeffbloo committed 5 years ago

Verified af7d4534

Add a few more things to the helper python scripts. (#3842)

skottmckay committed 5 years ago

Verified 0cdc1459

Update TensorRT dynamic shape profile when input shape changed during runtime (#3904)

stevenlix committed 5 years ago

Verified 28f693a8

add test to api tests to ensure correct hresult is returned for corrupt model file (#3796)

Ori Levari committed 5 years ago

Verified 6d2d9278

remove winrt (#3899)

smk2007 committed 5 years ago

Verified c7da1943

Added aarch64 build pipeline (#3841)

Prabhat committed 5 years ago

Verified ce3678ff

Use MlasComputeSoftmax in traditional ML ops (#3892)

skottmckay committed 5 years ago

Verified d7e39569

Stop proceeding with constant folding if a CPU kernel is not found found (#3836)

hariharans29 committed 5 years ago

Verified 06985a99

Add signed GPU nuget package to publish ort-nightly nuget feed (#3834)

Ryan Lai committed 5 years ago

Verified 7fd2c8f9

Update ONNX submodule to ONNX 1.7 release branch. (#3888)

codemzs committed 5 years ago

Verified 5e1244eb

dashboard integration - output training perf metrics as json (#3809)

ytaous committed 5 years ago

Verified 96030fdc

Re-enable Dropout(12) ONNX inferencing tests. (#3833)

codemzs committed 5 years ago

Verified 8291065e

Do not register Dropout(12) as training ONLY kernel. (#3859)

codemzs committed 5 years ago

Verified eb33d5ed

Load provider shared libraries relative to core runtime executable (#3884)

RyanUnderhill committed 5 years ago

Verified 408f62dd

Fold Shape Node During Constant Folding (#3748)

Lafi7e committed 5 years ago

Verified 3c248415

winml_is_inbox is not accessible in function scope, move it out. (#3886)

smk2007 committed 5 years ago

Verified 890c945d

fix dnnl ep shared lib python packaging. (#3875)

jywu-msft committed 5 years ago

Verified 672e40ba

fix build break on Linux. (#3874)

jywu-msft committed 5 years ago

Verified 93e39b81

Fix Windows Inbox build failing on 1) building raw api tests and 2) referencing _winml namespace in onnxruntime.dll (#3872)

smk2007 committed 5 years ago

Verified cf6a1c17

Fix C# log APIs. Also fixes github issue #3409. (#3840)

Pranav Sharma committed 5 years ago

Verified 22a71145

Newer Older