microsoft/onnxruntime

Pull Requests Commits

fix quantization of EmbeddingLayerNorm (#5321)

yufenglee committed 5 years ago

Verified 5de47aff

Cherry pick (batch 2) to rel-1.5.1 (#5290)

tianleiwu committed 5 years ago

Verified c00e13a2

Handle missing initializers in allocation planner to fix crashes with DML provider (#5244)

jeffbloo committed 5 years ago

389cca7a

ORT DirectML EP for Iron release, ONNX 1.5 (part 2) (#5263)

fdwr committed 5 years ago

b648fe5f

Fix bug in the back to back quantization of matmul and conv (#5264)

yufenglee committed 5 years ago

eb75b492

bump version to 1.5.1 (#5258)

tianleiwu committed 5 years ago

47447da4

Fix reshape fusion crash (#5252)

wangyems committed 5 years ago

87b15f32

Fix possible ios build break after update to Xcode 12 (#5246)

guoyu-wang committed 5 years ago

fc259de3

Place Shape's output in CPU memory (#5245)

SherlockNoMad committed 5 years ago

9fd76c86

Update BUILD.md training dependency info. (#5240)

edgchen1 committed 5 years ago

9158679c

Update BUILD.md for CUDA versions (#5239)

Changming Sun committed 5 years ago

b9b7c279

update TensorRT docs (#5238)

jywu-msft committed 5 years ago

0cbe240e

Revert to using release SafeInt repo now that it supports a build with exceptions disabled. (#5233)

skottmckay committed 5 years ago

c93f292d

Fix TransposeScaleMatMul and MatMulScaleFusion issues (#5230)

edgchen1 committed 5 years ago

6371ad61

Create profile for all dynamic shape input tensors (#5229)

stevenlix committed 5 years ago

c27f461c

[java] Fixing the buffer semantics. (#5223)

Craigacp committed 5 years ago

4427b1e2

fix _WIN32 (#5218)

jywu-msft committed 5 years ago

c909c677

Update conversion script and process to simplify creating ORT format models and a minimal build (#5217)

skottmckay committed 5 years ago

95b2e316

--shm-size=1024m to fix nccl shared memory issue (#5214)

liqunfu committed 5 years ago

21a7afb2

Remove shape inference and fix save large model(>2g) issue (#5210)

YUNQIUGUO committed 5 years ago

b791402f

Fix order of returned values in quantize_weight_per_channel (#5205)

pranav-prakash committed 5 years ago

0a31b9ed

NCHWc optimizer fixes for quantized models (#5203)

tracysh committed 5 years ago

f726af34

Corrects doc typos and formatting (#5201)

smkarlap committed 5 years ago

84ffdbc4

Add API to allow configuration of the global thread pools. (#5199)

Pranav Sharma committed 5 years ago

24d111c3

MaxPool versioning in quantization tools. (#5194)

zhanghuanrong committed 5 years ago

498483b4

Fix softmax_warp_backward math when is_log_softmax = True and register LogSoftmax CUDA kernel (#5160)

Suffian Khan committed 5 years ago

39a7f96a

Amdmigraphx improvements (#5158)

scxiao committed 5 years ago

8e650c53

Fix a bug in EmbedLayerNorm fusion (#5150)

wangyems committed 5 years ago

b693cb13

Update MCR CUDA docker image to 10.2 (#5181)

Changming Sun committed 5 years ago

5b5bcba9

Refactor TensorAt, prepare for release (#5180)

yuslepukhin committed 5 years ago

ece9a7c1

Older