strisunshinewentingwang/pytorch

Pull Requests Commits

Use scaled_dot_product_attention within attention.cpp (#87312)

drisspg committed 3 years ago

e24ce484

Add sequence number support for UCC (#85047)

Fuzzkatt committed 3 years ago

d13f1e6a

[ONNX] Fix get wrong summary of the docstring in `torch.onnx._deprecation.deprecated` (#87194)

HAOCHENYE committed 3 years ago

9642a7c2

[dynamo][dashboard] minor fixes for a clean Dashboard (#88056)

anijain2305 committed 3 years ago

d67b2ede

Even "nvcc not found" should be commented out (#87959)

brad-mengchi committed 3 years ago

9109ecf9

[dynamo][benchmarks] use fresh inductor cache and raise batch size wherever possible (#88044)

anijain2305 committed 3 years ago

1b575782

[BE] Do not package caffe2 in wheel (#87986)

malfet committed 3 years ago

e7b854fa

[vision hash update] update the pinned vision hash (#87948)

pytorchmergebot committed 3 years ago

65e77195

[BE] Do not assign string literal to `char *` (#87949)

malfet committed 3 years ago

621158cd

[Inductor] Enable Inductor unspec inputs test for different dtypes (#87809)

yanboliang committed 3 years ago

59001d05

Revert "Unify meta tensor and fake tensor converter conversion (#87943)"

pytorchmergebot committed 3 years ago

bc64999b

torchdynamo and xla integration (#87741)

shunting314 committed 3 years ago

e4a8661a

Use TORCH_CHECK instead of inappropriate CUDA_KERNEL_ASSERT (#87714)

r-barnes committed 3 years ago

6cd25eb6

[BE] Upload GHA artifacts to S3 (#87827)

huydhn committed 3 years ago

384b84d6

Add composable activation checkpointing (#87664)

mrshenli committed 3 years ago

d9b6e41d

Make barrier blocking in UCC (#86961)

Sergei-Lebedev committed 3 years ago

19171a21

Unify meta tensor and fake tensor converter conversion (#87943)

ezyang committed 3 years ago

baa715e7

[ONNX] Add internal node kind parsing (#87638)

titaiwangms committed 3 years ago

4210cebc

[ONNX] Parametrized Avgpool2D test to have all test combinations (#87893)

titaiwangms committed 3 years ago

cb05a4da

[ONNX] Disable ONNX ceil_mode and count_include_pad to aligntorch ceil_mode results in corner case (#87892)

titaiwangms committed 3 years ago

f2ae4593

Cleanup macos common conda installation (#87816)

huydhn committed 3 years ago

c810489d

Store usage log on GitHub when S3 is not available (#87947)

huydhn committed 3 years ago

53fea905

Fix pybind11 problems with c10::SymInt unregistered (#88011)

ezyang committed 3 years ago

d3c01c72

[FSDP()][2/N] Refactor training state (#87916)

Andrew Gu committed 3 years ago

e667c006

[FSDP()][1/N] Start refactoring FSDP root pre-forward (#87915)

Andrew Gu committed 3 years ago

cbc9faeb

Revert "[ONNX] Deprecate operators.py (#87798)"

pytorchmergebot committed 3 years ago

edd6cf99

[ONNX] Move all torch.onnx.export related tests to test/onnx (#87292)

titaiwangms committed 3 years ago

e3e84830

Move workspace related functions to separate file (#87651)

BrandonTheBuilder committed 3 years ago

1dad051b

[C10D][BE] Add exception handlers to c10d collectives function (#87643) (#87988)

wz337 committed 3 years ago

0cf572ff

Allow caffe2 to build with fbcode/mode/mac (#87293)

TovlyFB committed 3 years ago

20e16c01

Newer Older