strisunshinewentingwang/pytorch

Pull Requests Commits

[MPS] Add TORCH_CHECK for Convolution (#95495)

kulinseth committed 1 year ago

Verified 4e4d4b0a

Reserve the tensorrt backend name for torch-tensorrt (#95627)

Wei committed 1 year ago

Verified c4fa8508

Add float to list of allowed ops (#94910) (#95661)

atalman committed 1 year ago

Verified 36ead098

fix spurious aot autograd warning (#95521) (#95614)

bdhirsh committed 1 year ago

Verified 66d23dba

[CUDA][CUBLAS] Explicitly link against `cuBLASLt` (#95094) (#95615)

atalman committed 1 year ago

Verified e2fff588

Update triton hash (#95540) (#95577)

weiwangmeta committed 1 year ago

Verified 735333a7

[MPS] LSTM fixes (#95388)

kulinseth committed 1 year ago

Verified 60174888

[optim] Add general documentation on our algorithm defaults (#95391) (#95516)

janeyx99 committed 1 year ago

Verified e51e5e72

hotfix for memory leak in aot autograd induced by saving tensors for backward (#95101) (#95477)

bdhirsh committed 1 year ago

Verified 91739a02

inductor: fix complier error when trying to vectorize logit_and and logit_or (#95361) (#95439)

XiaobingSuper committed 1 year ago

Verified 531f097b

[optim] Set defaults to foreach, NOT fused (#95241) (#95415)

janeyx99 committed 1 year ago

Verified 00eb7b0d

[SDPA] Fix bug in parsing scaled_dot_product_attention arguments (#95311) (#95397)

drisspg committed 1 year ago

Verified 2180f342

use 4 warps for small block config in mm (#95383)

ngimel committed 1 year ago

Verified a90b4f09

[MPS] Fix issues with max_pool2d (#95325)

kulinseth committed 1 year ago

Verified 1211ceea

[MPS] View fixes (#95323)

kulinseth committed 1 year ago

Verified beaa5c59

Fix warning if backend registers timer (#91702) (#95363)

jeejakp12 committed 1 year ago

Verified 4bd5c1e4

Raise error on 3.11 dynamo export (#95088) (#95396)

williamwen42 committed 1 year ago

Verified f3c97a4e

[MPS] Copy fixes for MPS backend (#95321)

kulinseth committed 1 year ago

Verified 30cf0e70

[MPS] Fixes in backward functions of the MPS ops (#95327)

kulinseth committed 1 year ago

Verified 96f627dc

[MPS] Convolution fixes (#95318)

kulinseth committed 1 year ago

Verified 6f11e6d6

[MPS] Numerical stability and reduction fixes (#95317)

kulinseth committed 1 year ago

Verified fcec27f7

Raise error if torch.compile is called from windows or py 3.11 (#94940) (#95329)

atalman committed 1 year ago

Verified cddcb1e5

[profiler] update docs with repeat=1 (#95085) (#95242)

davidberard98 committed 1 year ago

Verified 0553b46d

fix numpy1.24 deprecations in unittests (#93997) (#95150)

ppwwyyxx committed 1 year ago

Verified b45d7697

Revert "[CI] Use prebuilt triton from nightly repo (#94732)" (#95310)

atalman committed 1 year ago

Verified 7ebb3094

Upgrade setuptools before building wheels (#95288)

atalman committed 1 year ago

Verified cedfcdab

Update triton hash (#95247) (#95285)

atalman committed 1 year ago

Verified 0b21e624

Deprecate Caffe2 ONNX exporter (#95071)

Thiago Crepaldi committed 1 year ago

Verified 91994c99

inductor(cpu): fix C++ compile error when sigmoid's post ops is a reduction op (#94890) (#95054)

XiaobingSuper committed 1 year ago

Verified 0b82f588

fix performance issue in torch.sparse.mm reduce mode (#94969) (#95018)

mingfeima committed 1 year ago

Verified 1f7ab1c8

Newer Older