microsoft/onnxruntime

Pull Requests Commits

expose optimized_model_filepath

ryotatomioka committed 5 years ago

6be3819e

Don't mark issues that are marked as enhancement as stale (#6134)

Pranav Sharma committed 5 years ago

Verified a9548283

Reduce amount of compiled CUDA device code (#6118)

edgchen1 committed 5 years ago

Verified 9810b9e0

Enable C# .NET5 for WinML (#6120)

smk2007 committed 5 years ago

Verified a6a23db1

Unify activation and initializer alignment value (#6109)

SherlockNoMad committed 5 years ago

Verified eb5c1f0f

Liqun/move nightly pl to linux multi gpu v100 (#6024)

liqunfu committed 5 years ago

Verified cde723a1

state_dict and load_state_dict for ORTTrainer (#6095)

baijumeswani committed 5 years ago

Verified dd2e5a1a

Bump ini from 1.3.5 to 1.3.8 in /nodejs

dependabot[bot] committed 5 years ago

d4dddd99

Honor auto_pad attribute in ConvTranspose (#4271)

hariharans29 committed 5 years ago

Verified c755ca0b

Fix multi-tensor LAMB reduction to be deterministic (#6028)

Suffian Khan committed 5 years ago

Verified 6cb5d3ac

Fix DEBUG_NODE_INPUTS_OUTPUTS test by putting it in a separate process, clean up unused test_main.cc files. (#5949)

edgchen1 committed 5 years ago

Verified c8ac34d6

Introduce VariadicAlias, remove hardcoded alias limits (#6106)

SherlockNoMad committed 5 years ago

Verified a53f4dd3

Make ROCM and CUDA reduction_all code more similar.

jessebenson committed 5 years ago

38c49c24

Implement conversion from ORT String to WinML Tensor String (#6097)

Ryan Lai committed 5 years ago

Verified 1eb146f5

Add skip test reason for onnx model zoo models and tier 2 models (#6081)

Ryan Lai committed 5 years ago

Verified 8bcb5fd1

If building inbox, hook up winrt_activation_handler for WinML Tests (#6074)

Ryan Lai committed 5 years ago

Verified 753af576

adding fp16 support for topk cuda kernel (#6082)

duli2012 committed 5 years ago

Verified e945b5fc

Add ReduceL2Grad and ClipGrad (#5970)

Lafi7e committed 5 years ago

Verified 7ddeafdf

Enable varied input type for custom op (#6066)

RandySheriffH committed 5 years ago

Verified 404982de

Update AMD transpose to match CUDA transpose.

jessebenson committed 5 years ago

cc47cfcb

Reduction kernel optimization (#6088)

edgchen1 committed 5 years ago

Verified abdbb5fc

Deprecate opsets <12 for training. (#6027)

Sergii Dymchenko committed 5 years ago

Verified 9e26e59a

clean un-used code. (#6059)

weixingzhang committed 5 years ago

Verified d95fc5e8

add dockerfile for ROCm3.10 and update BUILD.md for ROCm EP (#5821)

weixingzhang committed 5 years ago

Verified 27051157

Enable passing initial optimizer state while creating training session (#5869)

ashbhandare committed 5 years ago

Verified b1a75d0e

Fix AllReduce kernel for contiguous buffer (#6064)

SherlockNoMad committed 5 years ago

Verified 7a43fa00

Fix build definition template typo, add logging (#6065)

edgchen1 committed 5 years ago

Verified e3574867

save data to and load data from an hdf5 file for checkpointing (#5975)

baijumeswani committed 5 years ago

Verified 523d1871

Update version to 1.6.0 (#6041)

duli2012 committed 5 years ago

Verified 3e81711a

Android code coverage (#6061)

satyajandhyala committed 5 years ago

Verified f68a2561

Older