strisunshinewentingwang/pytorch

Pull Requests Commits

Fix flaky Dynamo export tests (#96488)

ezyang committed 1 year ago

80ce1a93

[Dynamo] Support torch.{cuda/cpu}.amp.autocast (#95416)

yanboliang committed 1 year ago

7fcf8b18

Require DOCTEST_SHOW environ to run plt.show (#96522)

Erotemic committed 1 year ago

d05f2ae4

[ONNX] Preserve stacktrace info for decomp (#95929)

BowenBao committed 1 year ago

384545bf

[ONNX] Move graph transform functions to 'passes' (#95664)

BowenBao committed 1 year ago

b97ce377

Revert "[PyTorch] Use c10::FastMap for memoizing in Pickler (#96360)" (#96547)

izaitsevfb committed 1 year ago

41991710

[BE][MPS] Use convenience functions (#96521)

malfet committed 1 year ago

42909114

Fix broken anchor in RELEASE.md (#96525)

malfet committed 1 year ago

85961f57

Reduce pytest blocklist part 2 (#96397)

clee2000 committed 1 year ago

4519228f

[Inductor Perf CI] Lower the threshold of performance smoke test speedup. (#96531)

weiwangmeta committed 1 year ago

49eed50d

[Profiler] Add export_memory_timeline to save memory timeline plot to file (#96137)

aaronenyeshi committed 1 year ago

e948ba07

[CI] handle more dynamo benchmark models that are not expected to be deterministic (#96324)

davidberard98 committed 1 year ago

29cd60df

Remove land checks in trymerge (#96401)

clee2000 committed 1 year ago

481582eb

Remove on_green and mandatory_only (#96400)

clee2000 committed 1 year ago

6894bb0a

[QOL] Raise a NameError when accessing non-existent variable (#96418)

ezyang committed 1 year ago

219d5eb4

add/add_ for sparse compressed formats: fix silent index downcast int64 -> int32 (#95294)

nikitaved committed 1 year ago

55cf7eef

[CI] Change compile_threads to 1 when running benchmark accuracy test on CI (#96195)

desertfire committed 1 year ago

a651e625

[DataPipe] Add `copy` option to `fork` DataPipe (#96030)

falckt committed 1 year ago

939c4ae6

Retry XLA dependencies installation step (#96352)

huydhn committed 1 year ago

e35f0204

[SPMD] Add defunctionalize_optimizer feature (#96323)

fegin committed 1 year ago

55d4842a

Switch AsyncCollectiveTensor to be a wrapper subclass. (#96105)

Rodrigo Kumpera committed 1 year ago

c7bd9b94

Enable thp(transparent huge pages) for buffer sizes >=2MB (#95963)

snadampal committed 1 year ago

3bb16a08

[XPU][Profiler] Add API support for XPU profiler to Kineto path (#94502)

xunsongh committed 1 year ago

b053a0f2

flatten_indices: remove syncs (#94401)

nikitaved committed 1 year ago

d0f4d629

Add PyObjectSlot member to StorageImpl (#93342)

kurtamohler committed 1 year ago

1b59c3fe

[fix] resize_ and resize_as_ : version bump (#96403)

kshitij12345 committed 1 year ago

987eade3

[caffe2] dont call cudnnDestroy on thread exit (crashes on windows with cuda 11/12) (#95382)

fuzic committed 1 year ago

8bce88d9

new triton main pin (#95896)

ngimel committed 1 year ago

76cac709

reland #96249: [inductor] show more kernel specific metrics in the benchmark result (#96461)

shunting314 committed 1 year ago

9aa216cb

Revert "new triton main pin (#95896)"

pytorchmergebot committed 1 year ago

d0731271

Newer Older