strisunshinewentingwang/pytorch

Pull Requests Commits

remove dynamo hash updates

clee2000 committed 2 years ago

e1dd23c7

[functorch] Fix cross to match unbatched behavior (#86926)

samdow committed 2 years ago

f3969bd8

Avoid calling logging.basicConfig (#86959)

SherlockNoMad committed 2 years ago

e271e823

Add meta support for _adaptive_avg_pool2d_backward (#86359) (#87074)

anjali411 committed 2 years ago

63512205

Revert "[Dynamo] More robust pyop support, module properties as args (#87020)"

pytorchmergebot committed 2 years ago

66715767

fix cudagraphify for inplace parameter change (#87060)

Natalia Gimelshein committed 2 years ago

8617f5f4

Revert "[inductor] Use decomps for unfold (#87025)"

pytorchmergebot committed 2 years ago

2c6167c4

[inductor] Set correct strides in fallback example run (#87049)

anijain2305 committed 2 years ago

2b558138

ATen/native (2/6): Use per-operator headers (#75572)

peterbell10 committed 2 years ago

4e5357fa

conv backward impl (#87047)

albanD committed 2 years ago

b40f4434

autograd clone_obey_contract() symint support (#87044)

albanD committed 2 years ago

1463013c

meta funcs for avg_pool2d and avg_pool2d_backward (#87043)

albanD committed 2 years ago

86c2e44c

Very limited pow support (#87042)

albanD committed 2 years ago

c21dcffc

[xla hash update] update the pinned xla hash (#87067)

pytorchmergebot committed 2 years ago

37e9e89a

[primTorch] Add a ref for `narrow_copy` (#86748)

nkaretnikov committed 2 years ago

91b3cd0b

[primTorch] Implement NLL loss reference (#81128)

rdspring1 committed 2 years ago

847ded6d

[TorchInductor] enable inplace buffers by default (#87037)

jgong5 committed 2 years ago

78e22890

Make `AdamW`, `NAdam` & `RAdam` differentiable (#86183)

Emilio Castillo committed 2 years ago

1b43883f

[vision hash update] update the pinned vision hash (#87021)

pytorchmergebot committed 2 years ago

364a9973

Reland 3 of Merge more symbolic meta kernels and symint changes from branch (#86795)

albanD committed 2 years ago

3a4c0900

[inductor] Disable parallel compile (#87048)

jansel committed 2 years ago

0379af68

stft: Require return_complex to be passed explicitly for real input (#86724)

peterbell10 committed 2 years ago

3007efda

[torchdynamo] Use ProcessPoolExecutor for triton compiles (#87032)

zdevito committed 2 years ago

2b7236a0

Migrate dynamo CI test shards to torch._dynamo (#87039)

jansel committed 2 years ago

945d333a

[inductor] Move size asserts to C++, fix bug (#87028)

jansel committed 2 years ago

30f6f690

[dynamo] Put printing graph breaks behind a config option (#87026)

jansel committed 2 years ago

d45e99ac

OpInfo: Sample input cleanup (4/n) (#86324)

peterbell10 committed 2 years ago

2a6d37d2

[inductor] Use decomps for unfold (#87025)

jansel committed 2 years ago

5099883f

Add labeler with dynamo/inductor paths to start (#87024)

ezyang committed 2 years ago

8a8cd092

Add triton to CI (#86988)

jansel committed 2 years ago

a0c2a7f2

Older