microsoft/onnxruntime

Pull Requests Commits

[DML EP] Enable more MHA masks (#17882)

PatriceVignola committed 2 years ago

Verified 65575389

support for fp16 in GetClipMinMax() (#17967)

guschmue committed 2 years ago

Verified 2ef7abf1

split kv for making mha and att ops faster

aciddelgado committed 2 years ago

fffbb032

start work on split kv mha and att

aciddelgado committed 2 years ago

1993458e

[java] Make the backing byte buffer in an OrtValue accessible (#16578)

Craigacp committed 2 years ago

Verified 34568314

Move a nodejs test to a different machine pool (#17970)

Changming Sun committed 2 years ago

Verified 57c87365

Fix AMD builds and enable testing NHWC CUDA ops in one GPU CI (#17972)

hariharans29 committed 2 years ago

Verified 93569867

Fix missing attribute on C# DOrtGetResizedStringTensorElementBuffer delegate (#17901)

skottmckay committed 2 years ago

Verified 6832b688

[CUDA] Fix SkipLayerNorm vectorized kernel out-of-bounds read (#17943)

tianleiwu committed 2 years ago

Verified 83547d30

Bump @babel/traverse from 7.18.2 to 7.23.2 in /js/react_native (#17962)

dependabot[bot] committed 2 years ago

Verified cf974f09

Make CUDA a NHWC EP (#17200)

gedoensmax committed 2 years ago

Verified 7c17e33c

[TensorRT EP] Fix cmake install (#17923)

chilo-ms committed 2 years ago

Verified 8abaa7b7

[js/web] optimize tsc for web: split out "npm prepare" (#17955)

fs-eire committed 2 years ago

Verified ad817d0e

enable training for win-wasm-ci.yml (#17954)

fs-eire committed 2 years ago

Verified f7341e81

add onnxruntime_float16.h into header validation list. (#17935)

mszhanyi committed 2 years ago

Verified 209b6dbd

Attempt to make the usage of the Android emulator in CIs more robust (#17903)

skottmckay committed 2 years ago

Verified ae211999

Add save_attribute option to quantize_static (#17945)

zhipenghan committed 2 years ago

Verified a55b2688

Add MatMul 4bits support on GPU (#17890)

yufenglee committed 2 years ago

Verified 11af3444

[QNN EP] Fix index-out-of-bounds bug in Slice builder when initializer is shared (#17905)

adrianlizarraga committed 2 years ago

Verified 3b69d9b8

[WebNN EP] Fixed wasm heap overflow for big model (#17902)

Honry committed 2 years ago

Verified 9c65d555

Custom op shape inference API (#17737)

RandySheriffH committed 2 years ago

Verified c6c3555d

Support output cross qk, dtw and more for whisper model (#17500)

zhanghuanrong committed 2 years ago

Verified 762703e0

Update eval_squad to use API of latest optimum (#17918)

tianleiwu committed 2 years ago

Verified c695de91

Fix build break with cuda 12.2 (#17922)

yufenglee committed 2 years ago

Verified 7551dd03

[TensorRT EP] Add back creation cudnn and cublas using external cuda stream (#17912)

chilo-ms committed 2 years ago

Verified 28c19445

Remove half of the compiled shaders for gridsample with unused types (#17928)

smk2007 committed 2 years ago

Verified 635d3faa

[CUDA] Fix SkipLayerNorm strict mode when skip has broadcast (#17896)

tianleiwu committed 2 years ago

Verified 67d7eb3a

[QNN EP] Fix topological node unit traversal during validation (#17913)

adrianlizarraga committed 2 years ago

Verified 4e2a03d5

[QNN EP] Handle rank 3 InstanceNormalization with N != 1 (#17897)

adrianlizarraga committed 2 years ago

Verified dad70ad4

[ROCm] Add ROCm Debug wheels to private ADO Feeds (#17887)

PeixuanZuo committed 2 years ago

Verified 0c5b1598

Newer Older