strisunshinewentingwang/pytorch

Pull Requests Commits

Benchmarks: tweak PE config settings.

Mikhail Zolotukhin committed 3 years ago

4eeeee3d

CI: Add a run of FastRNN benchmarks in default executor/fuser configuration.

Mikhail Zolotukhin committed 3 years ago

6e704c76

Benchmarks: add 'default' options for fuser and executor.

Mikhail Zolotukhin committed 3 years ago

9b700c88

SyncBN: preserve qconfig if it exists (#45317)

vkuzo committed 3 years ago

bdf329ef

Revert D23841786: [pytorch][PR] Enable distributed package on windows, Gloo backend supported only

Mike Ruberry committed 3 years ago

103fa389

[quant] Remove unused qconfig argument in qat linear module (#45307)

jerryzh168 committed 3 years ago

bc3151de

[RFC] Remove per-op-registration related code in caffe2/tools/codegen/gen.py (#45134)

dhruvbird committed 3 years ago

31ae8117

Enable distributed package on windows, Gloo backend supported only (#42897)

gunandrose4u committed 3 years ago

0122299f

[reland] Make grad point to bucket buffer in DDP to save memory usage (#44344)

zhaojuanmao committed 3 years ago

c6500bcf

[pytorch] refine dispatch keys in native_functions.yaml (2/N) (#45284)

ljk53 committed 3 years ago

630bd85a

[minor] Fix undefined variable (#45246)

xwang233 committed 3 years ago

7e5492e1

log metadata when model loading failed (#44430)

linbinyu committed 3 years ago

0f2c648c

Resend diff D23858329 (#45315)

movefast1990 committed 3 years ago

03dde4c6

[aten] Call fbgemm functions for embedding prepack/unpack (#44845)

dskhudia committed 3 years ago

677a59dc

Resolve comments in #44354. (#45150)

Ailing Zhang committed 3 years ago

0b6e5ad4

added check for NumberType (#44375)

hlahkar committed 3 years ago

92ebb04f

Fix test_rpc_profiling_remote_record_function (#45162)

rohan-varma committed 3 years ago

bee1d448

[JIT] Regularize tensorexpr fuser strategy with other fusers (#44972)

Elias Ellison committed 3 years ago

5dd288eb

Refactor subgraph merging (#44238)

Elias Ellison committed 3 years ago

0137e364

Add operator to compute the equalization scale (#45096)

Haixin Liu committed 3 years ago

1539d4a6

Add architectural support for multi-GPU. (#44059)

Ashkan Aliabadi committed 3 years ago

5a593306

Minor touchups. (#44317)

Ashkan Aliabadi committed 3 years ago

6311c5a4

Grammatically updated the tech docs (#45192)

Vishal487 committed 3 years ago

b84dd771

[caffe2] adds hypothesis test for queue ops cancel (#45178)

dahsh committed 3 years ago

cd7a6822

[JIT] Specialize AutogradZero: merge AutogradAnyNonZero and Not(AutogradAnyNonZero) checks into one. (#44987)

Mikhail Zolotukhin committed 3 years ago

71e6ce66

[caffe2] adds Cancel to SafeDequeueBlobsOp and SafeEnqueueBlobsOp (#45177)

dahsh committed 3 years ago

cbe1eac1

Make ddp_comm_hook_wrapper a private method. (#44643)

Yi Wang committed 3 years ago

022ba5a7

[Caffe2] Fix LayerNormOp when batch_size == 0. (#45250)

xiaomengy committed 3 years ago

e2bcdc7b

Run pytorch_core CUDA tests on GPU using TPX

malfet committed 3 years ago

c3a5aed5

add rocm 3.8 to nightly builds (#45222)

jeffdaily committed 3 years ago

c211a910

Older