microsoft/onnxruntime

Pull Requests Commits

Training API to export the eval model to an inference model (#13345)

baijumeswani committed 3 years ago

Verified a46c599a

First round of fixes. (#13452)

jchen351 committed 3 years ago

Verified 8827c4bd

Add '$schema' entry to cgmanifest.json files. (#13444)

edgchen1 committed 3 years ago

Verified 601b74b9

Update tsaoptions.json: update the email alias (#13448)

Changming Sun committed 3 years ago

Verified 7d583322

Fix a PoliCheck finding in _hierarchical_ortmodule.py(#13462)

Lafi7e committed 3 years ago

Verified 805ec459

[DML EP] Don't fuse a capability outside the compile call (#13468)

sumitsays committed 3 years ago

Verified 490e4dde

Improve logging and default affinity mask generation (#13338)

yuslepukhin committed 3 years ago

Verified 1c8a22ec

revert cmake external file (#13459)

rui-ren committed 3 years ago

Verified 136e15bf

[EP-Perf-Dashboard] Decouple docker image name from branch name (#13449)

adrianlizarraga committed 3 years ago

Verified 8770201e

add more documents

Cheng Tang committed 3 years ago

a016a4f6

Fix warnings preventing Onnx build (#13447)

jpvillam-amd committed 3 years ago

Verified 48b2ec94

Add a script for quick profile analysis (#13423)

abudup committed 3 years ago

Verified 8fbdc6cc

Update SkipLayerNorm fusion rules (#13350)

PeixuanZuo committed 3 years ago

Verified a0cc289b

DML EP add einsum MatMul NHCW ops (#13440)

PatriceVignola committed 3 years ago

Verified ac48bdec

DML EP register all data types for Where operator (#13443)

PatriceVignola committed 3 years ago

Verified d5e8d592

fix training build

Cheng Tang committed 3 years ago

904acd51

add more documents

Cheng Tang committed 3 years ago

2ae3cb68

fix comments about execution context

Cheng Tang committed 3 years ago

419f2181

[ADD] fuse Matmul + fastgalu -> gemmfastgelu (#11699)

PeixuanZuo committed 3 years ago

Verified 70b73afd

add on device training to the packaging pipelines (#13446)

AdamLouly committed 3 years ago

Verified cf8bf0c1

turn stream into class

Cheng Tang committed 3 years ago

5f9dac80

Update Attention operator to support separated Q/K/V inputs (#13410)

tianleiwu committed 3 years ago

Verified 7aafd862

Move build machines with Nvidia M60 GPUs to Nvidia T4 (#13170)

Changming Sun committed 3 years ago

Verified a396a91c

Document generation for operator kernels, enable internal overload of DML EP to initialize on software-only devices (#13428)

fdwr committed 3 years ago

Verified 0201cd75

Add script for question answering (SQuAD) accuracy evaluation of BERT model (#12947)

tianleiwu committed 3 years ago

Verified d80212d4

Add Pre- and Post-tunning API to allow pre- and postprocessing of params (#13411)

cloudhan committed 3 years ago

Verified d82036db

[ORTModule] Add Env Variable to Control Disabling Custom AutoGrad Function Support (#13430)

Lafi7e committed 3 years ago

Verified b6a3562f

[NNAPI] Refactor `Resize` as layout insensitive (#13412)

wejoncy committed 3 years ago

Verified ea1bdb16

Exculde hipify option from policheck (#13431)

cloudhan committed 3 years ago

Verified 93f7a97a

[ROCm] Use SkipLayerNorm original implementation in kernel explorer (#13382)

PeixuanZuo committed 3 years ago

Verified 28f470c2

Newer Older