microsoft/onnxruntime

Pull Requests Commits

disable matmul scale fusing

chentaMS committed 5 years ago

ed0b5174

temp fix in layernorm

chentaMS committed 5 years ago

9a6b98c2

put layernorm's mean/var on fp32

chentaMS committed 5 years ago

6853b847

register part of cpu opset13 kernels; adjust reshapefusion order ; fix a bug in function impl

chentaMS committed 5 years ago

43fedcc9

temp rollback bert loss udpate

chentaMS committed 5 years ago

74cdb5c6

temp disable gpu fusing transformers

chentaMS committed 5 years ago

5c1648c4

support opset13 on transformers.

Lafi7e committed 5 years ago

a41694da

transform mixed precision before build gradient

Lafi7e committed 5 years ago

d4facec7

add telemetry for named dimension overrides (#4794)

Ori Levari committed 5 years ago

Verified 5899c119

Sahar/csharp support openvino (#4703)

sfatimar committed 5 years ago

Verified 0a0ac70e

Bump onnx to get bfloat16 in ops, and some update in ort to support bfloat16 (#4791)

souptc committed 5 years ago

Verified 1b1a6a4c

Support usage of C API with C++ standards older than C++11 (#4257)

bbugaev committed 5 years ago

Verified 8ba6b6a2

unattended-upgrades (#4804)

jywu-msft committed 5 years ago

Verified 8d2e2255

Make grad name unique (#4788)

ashbhandare committed 5 years ago

Verified 5a8962d3

Using cublasGemmBatchedEx/cublasGemmStridedBatchedEx for training (#4731)

weixingzhang committed 5 years ago

Verified afa89566

Eliminate redundant subexpressions (#3047)

MaximKalininMS committed 5 years ago

Verified ec36c793

C# samples: Faster R-CNN (#4733)

marcusturewicz committed 5 years ago

Verified ce65275e

Install AzureML support and commonly used packages in the training image. (#4790)

Sergii Dymchenko committed 5 years ago

Verified de268526

Fix bugs in TensorRT (#4780)

stevenlix committed 5 years ago

Verified 7acef875

enable build flag '--use_openmp' on MacOS (#4774)

fs-eire committed 5 years ago

Verified aa993e95

build fixes for https://github.com/microsoft/onnxruntime/pull/4721 (#4784)

jywu-msft committed 5 years ago

Verified f12e9de1

Java API: Documentation cleanup (#4395)

yuzawa-san committed 5 years ago

Verified aca34352

replace namespace reference with alias (#4786)

smk2007 committed 5 years ago

Verified 722602f3

Changes for BART dynamic shapes in reduction (#4730)

ashbhandare committed 5 years ago

Verified 5e7a6e78

Fix MatmulTransposeFusion::ApplyImpl() setting of modified flag (#4775)

edgchen1 committed 5 years ago

Verified 74b3b844

Refactor Cast to reduce binary size. (#4765)

skottmckay committed 5 years ago

Verified 8fb743f7

Honor allow_spinning at barrier at end of parallel sections (#4767)

tlh20 committed 5 years ago

Verified 9cec98ec

Update Python version support (#4778)

faxu committed 5 years ago

Verified 61b2a663

Add missing header file to MNIST.cpp (#4773)

Changming Sun committed 5 years ago

Verified cddddc4d

add --no_attention_mask option (#4750)

tianleiwu committed 5 years ago

Verified a69ca638

Older