microsoft/onnxruntime

Pull Requests Commits

[DML EP] Force instance norm inputs to be 4D to better target metacommands (#14020)

PatriceVignola committed 3 years ago

Verified e7f9d40d

[DML EP] Force layer norm inputs to be 4D to better target metacommands (#14022)

PatriceVignola committed 3 years ago

Verified 58961210

CloudEP (#13855)

RandySheriffH committed 3 years ago

Verified 587e891c

add dnnl dependency in unittest.cmake (#14104)

mszhanyi committed 3 years ago

Verified 52e3fe96

Remove all usages of hipLaunchKernelGGL (#14089)

cloudhan committed 3 years ago

Verified 613920d6

[CUDA] Update fused MHA to support flash attention and causal mask (#13953)

tianleiwu committed 3 years ago

Verified 6a9dc6c9

Support loading widechar paths on windows (#14066)

yuslepukhin committed 3 years ago

Verified 5d729839

Fix nightly ort training ci pipeline (#14007)

baijumeswani committed 3 years ago

Verified b8587895

fix the cuda EP test failure (#14087)

souptc committed 3 years ago

Verified 9f52a8bc

Fix prefast scan bug 8995 (#14077)

chenfucn committed 3 years ago

Verified a79d88ed

[ORTModule] ATen upsample_nearest Gradient Bugfix (#14069)

Lafi7e committed 3 years ago

Verified 0c3480e5

[Fix] CPU memory type need device_id=0 to get allocator (#14050)

PeixuanZuo committed 3 years ago

Verified 770fb964

Let Cmake decide where to place abseil (#14057)

yuslepukhin committed 3 years ago

Verified d762aa2a

Support for custom op variadic inputs/outputs (#13946)

adrianlizarraga committed 3 years ago

Verified 3bbcc279

Deprecate one step beam search (#14046)

tianleiwu committed 3 years ago

Verified 8ac264b8

expose lr scheduler python bindings for on device training. (#13882)

AdamLouly committed 3 years ago

Verified e49f3586

disable some test input for generation model temporarily (#14045)

gh-yewang committed 3 years ago

Verified b999022b

Sampling op (#13426)

gh-yewang committed 3 years ago

Verified 68518a1b

Replace all numpy.bool by python builtin bool (#14014)

fxmarty committed 3 years ago

Verified 4d2dc8bb

[QAT] Graph transformer to fuse QDQ pattern into FakeQuant (#13777)

baijumeswani committed 3 years ago

Verified 1b58331f

Support two stages onnx GPT-2 conversion (#14025)

tianleiwu committed 3 years ago

Verified 944bff0a

[ROCm] update skip_layernorm test sample (#14051)

PeixuanZuo committed 3 years ago

Verified 694ba033

Optimize computation orders (#13672)

pengwa committed 3 years ago

Verified 2f5bf75e

Support (Bias)SkipLayerNormalization fusion in GPT2 (#13988)

hariharans29 committed 3 years ago

Verified 7ed8bd4f

Add provider selection for gpt2/convert_to_onnx.py (#13982)

groenenboomj committed 3 years ago

Verified baba312e

[ROCm] Update Dockerfiles of ROCm and MIgraphX to ROCm5.4 (#14013)

PeixuanZuo committed 3 years ago

Verified a170e40f

[ROCm] Add ROCm5.4 to python package pipeline (#14012)

PeixuanZuo committed 3 years ago

Verified b5fd2a6a

[ROCm] Update ROCm and MigraphX CI to ROCm5.4 (#14011)

PeixuanZuo committed 3 years ago

Verified ab2dd8df

Update CUDA ArgMin/ArgMax op kernels to have end version 11 since opset 12+ is not supported yet. (#13983)

edgchen1 committed 3 years ago

Verified df8ff34f

DML EP Register operators for Opset 16 (#14034)

nums11 committed 3 years ago

Verified 8943d623

Newer Older