microsoft/onnxruntime

Pull Requests Commits

review comments

ashbhandare committed 5 years ago

6d86d4d6

Review comments

ashbhandare committed 5 years ago

132d3f02

Review comments

ashbhandare committed 5 years ago

eb3bf589

Addressing review comments

ashbhandare committed 5 years ago

59fb6762

add condition to use custom kernel for fp16 only for GPU with compute capability 7. or higher due to use of _half2

ashbhandare committed 5 years ago

0922dd11

Windows build fix

ashbhandare committed 5 years ago

cdf74d5e

Kernel changes, added fp32 and fp16 tests, removed redundant fwd kernel code.

ashbhandare committed 5 years ago

82bbadb0

Functionally working changes

ashbhandare committed 5 years ago

ae8d8b1b

Build new AVX file with /ARCH:AVX (#4442)

tracysh committed 5 years ago

Verified aa06d308

Remove use of RTTI in CUDA provider (#4444)

tiagoshibata committed 5 years ago

Verified e62686c3

Add cppwinrt and cswinrt tests in windowsai nuget pipeline (#4381)

smk2007 committed 5 years ago

Verified fdb4a3a2

add bias for DynamicQuantizeMatmul (#4440)

yufenglee committed 5 years ago

Verified 612f52c9

Update dependency introduced by fuzzing change. (#4438)

Pranav Sharma committed 5 years ago

Verified 1f1384f8

Add Fusion for GPT Attention with both past state and attention mask (#4437)

tianleiwu committed 5 years ago

Verified eabf6dc9

Change the input to NNAPI EP ModelBuilder from ModelProto to GraphViewer (#4389)

guoyu-wang committed 5 years ago

Verified 7baf3749

Onnxruntime fuzzing (#4341)

EronsJ committed 5 years ago

Verified 632b2896

Remove unused initializer in graph after embed fusion (#4436)

liuziyue committed 5 years ago

Verified ec35a1b5

MLAS: support prepacking APIs for quantized GEMM (#4433)

tracysh committed 5 years ago

Verified 3ef44981

add function initialization back to graph resolve (#4434)

askhade committed 5 years ago

Verified dd73e8c0

Liqun/roberta (#4408)

liqunfu committed 5 years ago

Verified 0fdb1e9f

use system libnsync (#4377)

mslacken committed 5 years ago

Verified 35884843

Fix symbolic_shape_infer for Resize with roi (#4426)

KeDengMS committed 5 years ago

Verified 77cf51b1

Fix Vitis-AI EP for memory info into IAllocator move (#4404)

jornt-xilinx committed 5 years ago

Verified 0d4a65ee

Optimize GatherND (#4097)

pengwa committed 5 years ago

Verified 8bcdefc9

Optimize LayernormGrad (#4156)

weixingzhang committed 5 years ago

Verified bd11ab68

optimize transpose CUDA kernel (#4233)

weixingzhang committed 5 years ago

Verified 33e06be4

Update BiasGeluGradDxKernel and tests. (#4400)

edgchen1 committed 5 years ago

Verified dba22b17

Use single OpKernel for u8u8 and u8s8 types (#4414)

tracysh committed 5 years ago

Verified 93d49647

Use the file size while reading onnx models. Ensure models are loaded using APIs in model.h for consistency. (#4399)

Pranav Sharma committed 5 years ago

Verified 4df8a1e2

Add ability to specify just the device when using IOBinding for an output (#4386)

skottmckay committed 5 years ago

Verified d22f6fdd

Older